{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 4810,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010395010395010396,
      "grad_norm": 16.988855361938477,
      "learning_rate": 0.0,
      "loss": 2.202,
      "step": 1
    },
    {
      "epoch": 0.002079002079002079,
      "grad_norm": 32.9708137512207,
      "learning_rate": 2.0790020790020793e-08,
      "loss": 2.4915,
      "step": 2
    },
    {
      "epoch": 0.0031185031185031187,
      "grad_norm": 19.217676162719727,
      "learning_rate": 4.158004158004159e-08,
      "loss": 2.2768,
      "step": 3
    },
    {
      "epoch": 0.004158004158004158,
      "grad_norm": 28.554401397705078,
      "learning_rate": 6.237006237006238e-08,
      "loss": 2.7587,
      "step": 4
    },
    {
      "epoch": 0.005197505197505198,
      "grad_norm": 16.13688087463379,
      "learning_rate": 8.316008316008317e-08,
      "loss": 2.0513,
      "step": 5
    },
    {
      "epoch": 0.006237006237006237,
      "grad_norm": 12.732123374938965,
      "learning_rate": 1.0395010395010397e-07,
      "loss": 1.8028,
      "step": 6
    },
    {
      "epoch": 0.007276507276507277,
      "grad_norm": 9.309935569763184,
      "learning_rate": 1.2474012474012475e-07,
      "loss": 1.9257,
      "step": 7
    },
    {
      "epoch": 0.008316008316008316,
      "grad_norm": 64.55175018310547,
      "learning_rate": 1.4553014553014554e-07,
      "loss": 2.121,
      "step": 8
    },
    {
      "epoch": 0.009355509355509356,
      "grad_norm": 21.101022720336914,
      "learning_rate": 1.6632016632016635e-07,
      "loss": 2.04,
      "step": 9
    },
    {
      "epoch": 0.010395010395010396,
      "grad_norm": 16.773162841796875,
      "learning_rate": 1.8711018711018713e-07,
      "loss": 2.189,
      "step": 10
    },
    {
      "epoch": 0.011434511434511435,
      "grad_norm": 18.621702194213867,
      "learning_rate": 2.0790020790020794e-07,
      "loss": 1.6861,
      "step": 11
    },
    {
      "epoch": 0.012474012474012475,
      "grad_norm": 19.770313262939453,
      "learning_rate": 2.2869022869022872e-07,
      "loss": 2.0745,
      "step": 12
    },
    {
      "epoch": 0.013513513513513514,
      "grad_norm": 9.97078800201416,
      "learning_rate": 2.494802494802495e-07,
      "loss": 1.7891,
      "step": 13
    },
    {
      "epoch": 0.014553014553014554,
      "grad_norm": 20.687963485717773,
      "learning_rate": 2.702702702702703e-07,
      "loss": 1.6521,
      "step": 14
    },
    {
      "epoch": 0.015592515592515593,
      "grad_norm": 11.27476692199707,
      "learning_rate": 2.910602910602911e-07,
      "loss": 1.6738,
      "step": 15
    },
    {
      "epoch": 0.016632016632016633,
      "grad_norm": 26.307353973388672,
      "learning_rate": 3.118503118503119e-07,
      "loss": 2.2328,
      "step": 16
    },
    {
      "epoch": 0.017671517671517672,
      "grad_norm": 15.594326972961426,
      "learning_rate": 3.326403326403327e-07,
      "loss": 2.0531,
      "step": 17
    },
    {
      "epoch": 0.018711018711018712,
      "grad_norm": 18.295095443725586,
      "learning_rate": 3.534303534303535e-07,
      "loss": 2.1244,
      "step": 18
    },
    {
      "epoch": 0.01975051975051975,
      "grad_norm": 30.941444396972656,
      "learning_rate": 3.7422037422037426e-07,
      "loss": 2.0255,
      "step": 19
    },
    {
      "epoch": 0.02079002079002079,
      "grad_norm": 28.534343719482422,
      "learning_rate": 3.9501039501039504e-07,
      "loss": 2.1367,
      "step": 20
    },
    {
      "epoch": 0.02182952182952183,
      "grad_norm": 105.21300506591797,
      "learning_rate": 4.158004158004159e-07,
      "loss": 2.2866,
      "step": 21
    },
    {
      "epoch": 0.02286902286902287,
      "grad_norm": 16.125764846801758,
      "learning_rate": 4.3659043659043666e-07,
      "loss": 2.4157,
      "step": 22
    },
    {
      "epoch": 0.02390852390852391,
      "grad_norm": 16.123016357421875,
      "learning_rate": 4.5738045738045745e-07,
      "loss": 2.5256,
      "step": 23
    },
    {
      "epoch": 0.02494802494802495,
      "grad_norm": 15.839056968688965,
      "learning_rate": 4.781704781704783e-07,
      "loss": 1.7972,
      "step": 24
    },
    {
      "epoch": 0.02598752598752599,
      "grad_norm": 21.773963928222656,
      "learning_rate": 4.98960498960499e-07,
      "loss": 2.2897,
      "step": 25
    },
    {
      "epoch": 0.02702702702702703,
      "grad_norm": 16.242046356201172,
      "learning_rate": 5.197505197505199e-07,
      "loss": 2.0965,
      "step": 26
    },
    {
      "epoch": 0.028066528066528068,
      "grad_norm": 12.057516098022461,
      "learning_rate": 5.405405405405406e-07,
      "loss": 1.9348,
      "step": 27
    },
    {
      "epoch": 0.029106029106029108,
      "grad_norm": 48.11043930053711,
      "learning_rate": 5.613305613305614e-07,
      "loss": 1.8615,
      "step": 28
    },
    {
      "epoch": 0.030145530145530147,
      "grad_norm": 28.504892349243164,
      "learning_rate": 5.821205821205821e-07,
      "loss": 2.5755,
      "step": 29
    },
    {
      "epoch": 0.031185031185031187,
      "grad_norm": 12.763325691223145,
      "learning_rate": 6.02910602910603e-07,
      "loss": 1.9018,
      "step": 30
    },
    {
      "epoch": 0.032224532224532226,
      "grad_norm": 13.687501907348633,
      "learning_rate": 6.237006237006238e-07,
      "loss": 2.444,
      "step": 31
    },
    {
      "epoch": 0.033264033264033266,
      "grad_norm": 8.045001029968262,
      "learning_rate": 6.444906444906446e-07,
      "loss": 1.7589,
      "step": 32
    },
    {
      "epoch": 0.034303534303534305,
      "grad_norm": 19.066606521606445,
      "learning_rate": 6.652806652806654e-07,
      "loss": 2.3628,
      "step": 33
    },
    {
      "epoch": 0.035343035343035345,
      "grad_norm": 13.532938003540039,
      "learning_rate": 6.860706860706861e-07,
      "loss": 2.3399,
      "step": 34
    },
    {
      "epoch": 0.036382536382536385,
      "grad_norm": 14.592718124389648,
      "learning_rate": 7.06860706860707e-07,
      "loss": 1.8444,
      "step": 35
    },
    {
      "epoch": 0.037422037422037424,
      "grad_norm": 19.344562530517578,
      "learning_rate": 7.276507276507278e-07,
      "loss": 2.4186,
      "step": 36
    },
    {
      "epoch": 0.038461538461538464,
      "grad_norm": 23.072294235229492,
      "learning_rate": 7.484407484407485e-07,
      "loss": 2.3203,
      "step": 37
    },
    {
      "epoch": 0.0395010395010395,
      "grad_norm": 20.00788116455078,
      "learning_rate": 7.692307692307694e-07,
      "loss": 1.9721,
      "step": 38
    },
    {
      "epoch": 0.04054054054054054,
      "grad_norm": 13.704610824584961,
      "learning_rate": 7.900207900207901e-07,
      "loss": 1.9419,
      "step": 39
    },
    {
      "epoch": 0.04158004158004158,
      "grad_norm": 35.170108795166016,
      "learning_rate": 8.108108108108109e-07,
      "loss": 2.1189,
      "step": 40
    },
    {
      "epoch": 0.04261954261954262,
      "grad_norm": 13.251538276672363,
      "learning_rate": 8.316008316008318e-07,
      "loss": 2.2046,
      "step": 41
    },
    {
      "epoch": 0.04365904365904366,
      "grad_norm": 37.792728424072266,
      "learning_rate": 8.523908523908525e-07,
      "loss": 2.3078,
      "step": 42
    },
    {
      "epoch": 0.0446985446985447,
      "grad_norm": 19.703136444091797,
      "learning_rate": 8.731808731808733e-07,
      "loss": 2.1895,
      "step": 43
    },
    {
      "epoch": 0.04573804573804574,
      "grad_norm": 25.02670669555664,
      "learning_rate": 8.939708939708941e-07,
      "loss": 2.209,
      "step": 44
    },
    {
      "epoch": 0.04677754677754678,
      "grad_norm": 11.132152557373047,
      "learning_rate": 9.147609147609149e-07,
      "loss": 1.8703,
      "step": 45
    },
    {
      "epoch": 0.04781704781704782,
      "grad_norm": 10.692375183105469,
      "learning_rate": 9.355509355509356e-07,
      "loss": 1.967,
      "step": 46
    },
    {
      "epoch": 0.04885654885654886,
      "grad_norm": 18.32746696472168,
      "learning_rate": 9.563409563409566e-07,
      "loss": 2.353,
      "step": 47
    },
    {
      "epoch": 0.0498960498960499,
      "grad_norm": 23.04522705078125,
      "learning_rate": 9.771309771309773e-07,
      "loss": 2.1668,
      "step": 48
    },
    {
      "epoch": 0.05093555093555094,
      "grad_norm": 13.668973922729492,
      "learning_rate": 9.97920997920998e-07,
      "loss": 1.7651,
      "step": 49
    },
    {
      "epoch": 0.05197505197505198,
      "grad_norm": 11.134869575500488,
      "learning_rate": 1.0187110187110188e-06,
      "loss": 1.8238,
      "step": 50
    },
    {
      "epoch": 0.05301455301455302,
      "grad_norm": 24.748937606811523,
      "learning_rate": 1.0395010395010397e-06,
      "loss": 2.0508,
      "step": 51
    },
    {
      "epoch": 0.05405405405405406,
      "grad_norm": 21.49311065673828,
      "learning_rate": 1.0602910602910604e-06,
      "loss": 1.9374,
      "step": 52
    },
    {
      "epoch": 0.0550935550935551,
      "grad_norm": 31.98406410217285,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 2.3183,
      "step": 53
    },
    {
      "epoch": 0.056133056133056136,
      "grad_norm": 14.93075180053711,
      "learning_rate": 1.1018711018711021e-06,
      "loss": 1.918,
      "step": 54
    },
    {
      "epoch": 0.057172557172557176,
      "grad_norm": 22.455198287963867,
      "learning_rate": 1.1226611226611228e-06,
      "loss": 1.9401,
      "step": 55
    },
    {
      "epoch": 0.058212058212058215,
      "grad_norm": 11.494815826416016,
      "learning_rate": 1.1434511434511436e-06,
      "loss": 1.9454,
      "step": 56
    },
    {
      "epoch": 0.059251559251559255,
      "grad_norm": 15.472561836242676,
      "learning_rate": 1.1642411642411643e-06,
      "loss": 2.1118,
      "step": 57
    },
    {
      "epoch": 0.060291060291060294,
      "grad_norm": 15.340880393981934,
      "learning_rate": 1.1850311850311852e-06,
      "loss": 2.1039,
      "step": 58
    },
    {
      "epoch": 0.061330561330561334,
      "grad_norm": 53.742252349853516,
      "learning_rate": 1.205821205821206e-06,
      "loss": 2.7519,
      "step": 59
    },
    {
      "epoch": 0.062370062370062374,
      "grad_norm": 21.00668716430664,
      "learning_rate": 1.2266112266112267e-06,
      "loss": 2.4038,
      "step": 60
    },
    {
      "epoch": 0.06340956340956341,
      "grad_norm": 32.80131149291992,
      "learning_rate": 1.2474012474012476e-06,
      "loss": 2.5428,
      "step": 61
    },
    {
      "epoch": 0.06444906444906445,
      "grad_norm": 15.214197158813477,
      "learning_rate": 1.2681912681912684e-06,
      "loss": 2.0914,
      "step": 62
    },
    {
      "epoch": 0.06548856548856549,
      "grad_norm": 26.892454147338867,
      "learning_rate": 1.288981288981289e-06,
      "loss": 1.6108,
      "step": 63
    },
    {
      "epoch": 0.06652806652806653,
      "grad_norm": 13.933700561523438,
      "learning_rate": 1.30977130977131e-06,
      "loss": 1.6882,
      "step": 64
    },
    {
      "epoch": 0.06756756756756757,
      "grad_norm": 10.220669746398926,
      "learning_rate": 1.3305613305613308e-06,
      "loss": 1.798,
      "step": 65
    },
    {
      "epoch": 0.06860706860706861,
      "grad_norm": 21.804941177368164,
      "learning_rate": 1.3513513513513515e-06,
      "loss": 1.8424,
      "step": 66
    },
    {
      "epoch": 0.06964656964656965,
      "grad_norm": 23.250431060791016,
      "learning_rate": 1.3721413721413722e-06,
      "loss": 2.4271,
      "step": 67
    },
    {
      "epoch": 0.07068607068607069,
      "grad_norm": 11.683350563049316,
      "learning_rate": 1.3929313929313932e-06,
      "loss": 1.937,
      "step": 68
    },
    {
      "epoch": 0.07172557172557173,
      "grad_norm": 37.01130294799805,
      "learning_rate": 1.413721413721414e-06,
      "loss": 2.0699,
      "step": 69
    },
    {
      "epoch": 0.07276507276507277,
      "grad_norm": 12.468998908996582,
      "learning_rate": 1.4345114345114346e-06,
      "loss": 2.1341,
      "step": 70
    },
    {
      "epoch": 0.07380457380457381,
      "grad_norm": 15.133217811584473,
      "learning_rate": 1.4553014553014556e-06,
      "loss": 2.2831,
      "step": 71
    },
    {
      "epoch": 0.07484407484407485,
      "grad_norm": 16.24321174621582,
      "learning_rate": 1.4760914760914763e-06,
      "loss": 2.013,
      "step": 72
    },
    {
      "epoch": 0.07588357588357589,
      "grad_norm": 16.070463180541992,
      "learning_rate": 1.496881496881497e-06,
      "loss": 2.3446,
      "step": 73
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 16.41120719909668,
      "learning_rate": 1.5176715176715178e-06,
      "loss": 2.0695,
      "step": 74
    },
    {
      "epoch": 0.07796257796257797,
      "grad_norm": 14.749588012695312,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 1.9838,
      "step": 75
    },
    {
      "epoch": 0.079002079002079,
      "grad_norm": 12.910445213317871,
      "learning_rate": 1.5592515592515594e-06,
      "loss": 1.9391,
      "step": 76
    },
    {
      "epoch": 0.08004158004158005,
      "grad_norm": 19.614336013793945,
      "learning_rate": 1.5800415800415802e-06,
      "loss": 1.9071,
      "step": 77
    },
    {
      "epoch": 0.08108108108108109,
      "grad_norm": 31.232622146606445,
      "learning_rate": 1.6008316008316011e-06,
      "loss": 1.905,
      "step": 78
    },
    {
      "epoch": 0.08212058212058213,
      "grad_norm": 12.483946800231934,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 2.0306,
      "step": 79
    },
    {
      "epoch": 0.08316008316008316,
      "grad_norm": 12.690123558044434,
      "learning_rate": 1.6424116424116426e-06,
      "loss": 1.9655,
      "step": 80
    },
    {
      "epoch": 0.0841995841995842,
      "grad_norm": 13.518976211547852,
      "learning_rate": 1.6632016632016635e-06,
      "loss": 2.1174,
      "step": 81
    },
    {
      "epoch": 0.08523908523908524,
      "grad_norm": 19.20243263244629,
      "learning_rate": 1.6839916839916843e-06,
      "loss": 2.306,
      "step": 82
    },
    {
      "epoch": 0.08627858627858628,
      "grad_norm": 10.600049018859863,
      "learning_rate": 1.704781704781705e-06,
      "loss": 2.0379,
      "step": 83
    },
    {
      "epoch": 0.08731808731808732,
      "grad_norm": 6.616295337677002,
      "learning_rate": 1.7255717255717257e-06,
      "loss": 1.8567,
      "step": 84
    },
    {
      "epoch": 0.08835758835758836,
      "grad_norm": 9.49284553527832,
      "learning_rate": 1.7463617463617467e-06,
      "loss": 1.844,
      "step": 85
    },
    {
      "epoch": 0.0893970893970894,
      "grad_norm": 13.307085037231445,
      "learning_rate": 1.7671517671517674e-06,
      "loss": 1.9826,
      "step": 86
    },
    {
      "epoch": 0.09043659043659044,
      "grad_norm": 13.201308250427246,
      "learning_rate": 1.7879417879417881e-06,
      "loss": 2.1031,
      "step": 87
    },
    {
      "epoch": 0.09147609147609148,
      "grad_norm": 13.714674949645996,
      "learning_rate": 1.808731808731809e-06,
      "loss": 1.5715,
      "step": 88
    },
    {
      "epoch": 0.09251559251559252,
      "grad_norm": 14.892278671264648,
      "learning_rate": 1.8295218295218298e-06,
      "loss": 1.7279,
      "step": 89
    },
    {
      "epoch": 0.09355509355509356,
      "grad_norm": 7.037039279937744,
      "learning_rate": 1.8503118503118505e-06,
      "loss": 1.9976,
      "step": 90
    },
    {
      "epoch": 0.0945945945945946,
      "grad_norm": 14.051565170288086,
      "learning_rate": 1.8711018711018713e-06,
      "loss": 2.0268,
      "step": 91
    },
    {
      "epoch": 0.09563409563409564,
      "grad_norm": 7.723928928375244,
      "learning_rate": 1.8918918918918922e-06,
      "loss": 2.0009,
      "step": 92
    },
    {
      "epoch": 0.09667359667359668,
      "grad_norm": 14.955673217773438,
      "learning_rate": 1.912681912681913e-06,
      "loss": 1.9555,
      "step": 93
    },
    {
      "epoch": 0.09771309771309772,
      "grad_norm": 10.495637893676758,
      "learning_rate": 1.9334719334719337e-06,
      "loss": 2.0791,
      "step": 94
    },
    {
      "epoch": 0.09875259875259876,
      "grad_norm": 9.99687385559082,
      "learning_rate": 1.9542619542619546e-06,
      "loss": 1.8874,
      "step": 95
    },
    {
      "epoch": 0.0997920997920998,
      "grad_norm": 13.424286842346191,
      "learning_rate": 1.975051975051975e-06,
      "loss": 1.916,
      "step": 96
    },
    {
      "epoch": 0.10083160083160084,
      "grad_norm": 28.673757553100586,
      "learning_rate": 1.995841995841996e-06,
      "loss": 2.0923,
      "step": 97
    },
    {
      "epoch": 0.10187110187110188,
      "grad_norm": 12.696794509887695,
      "learning_rate": 2.016632016632017e-06,
      "loss": 2.1206,
      "step": 98
    },
    {
      "epoch": 0.10291060291060292,
      "grad_norm": 14.118012428283691,
      "learning_rate": 2.0374220374220375e-06,
      "loss": 1.9878,
      "step": 99
    },
    {
      "epoch": 0.10395010395010396,
      "grad_norm": 9.848310470581055,
      "learning_rate": 2.0582120582120585e-06,
      "loss": 1.8487,
      "step": 100
    },
    {
      "epoch": 0.104989604989605,
      "grad_norm": 10.640583038330078,
      "learning_rate": 2.0790020790020794e-06,
      "loss": 2.2879,
      "step": 101
    },
    {
      "epoch": 0.10602910602910603,
      "grad_norm": 13.701424598693848,
      "learning_rate": 2.0997920997921e-06,
      "loss": 1.8333,
      "step": 102
    },
    {
      "epoch": 0.10706860706860707,
      "grad_norm": 14.907086372375488,
      "learning_rate": 2.120582120582121e-06,
      "loss": 1.9786,
      "step": 103
    },
    {
      "epoch": 0.10810810810810811,
      "grad_norm": 11.484697341918945,
      "learning_rate": 2.141372141372142e-06,
      "loss": 1.9187,
      "step": 104
    },
    {
      "epoch": 0.10914760914760915,
      "grad_norm": 14.736360549926758,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 1.8816,
      "step": 105
    },
    {
      "epoch": 0.1101871101871102,
      "grad_norm": 9.888904571533203,
      "learning_rate": 2.1829521829521833e-06,
      "loss": 1.9339,
      "step": 106
    },
    {
      "epoch": 0.11122661122661123,
      "grad_norm": 18.789710998535156,
      "learning_rate": 2.2037422037422042e-06,
      "loss": 1.8726,
      "step": 107
    },
    {
      "epoch": 0.11226611226611227,
      "grad_norm": 18.137393951416016,
      "learning_rate": 2.2245322245322247e-06,
      "loss": 2.1816,
      "step": 108
    },
    {
      "epoch": 0.11330561330561331,
      "grad_norm": 11.418164253234863,
      "learning_rate": 2.2453222453222457e-06,
      "loss": 1.8386,
      "step": 109
    },
    {
      "epoch": 0.11434511434511435,
      "grad_norm": 12.0963716506958,
      "learning_rate": 2.2661122661122666e-06,
      "loss": 1.4687,
      "step": 110
    },
    {
      "epoch": 0.11538461538461539,
      "grad_norm": 8.479762077331543,
      "learning_rate": 2.286902286902287e-06,
      "loss": 1.8498,
      "step": 111
    },
    {
      "epoch": 0.11642411642411643,
      "grad_norm": 15.996792793273926,
      "learning_rate": 2.307692307692308e-06,
      "loss": 2.0195,
      "step": 112
    },
    {
      "epoch": 0.11746361746361747,
      "grad_norm": 17.23140525817871,
      "learning_rate": 2.3284823284823286e-06,
      "loss": 1.8239,
      "step": 113
    },
    {
      "epoch": 0.11850311850311851,
      "grad_norm": 12.582114219665527,
      "learning_rate": 2.3492723492723495e-06,
      "loss": 1.7674,
      "step": 114
    },
    {
      "epoch": 0.11954261954261955,
      "grad_norm": 15.431148529052734,
      "learning_rate": 2.3700623700623705e-06,
      "loss": 1.9072,
      "step": 115
    },
    {
      "epoch": 0.12058212058212059,
      "grad_norm": 17.711864471435547,
      "learning_rate": 2.390852390852391e-06,
      "loss": 2.1458,
      "step": 116
    },
    {
      "epoch": 0.12162162162162163,
      "grad_norm": 12.348552703857422,
      "learning_rate": 2.411642411642412e-06,
      "loss": 1.706,
      "step": 117
    },
    {
      "epoch": 0.12266112266112267,
      "grad_norm": 20.33272361755371,
      "learning_rate": 2.432432432432433e-06,
      "loss": 1.6389,
      "step": 118
    },
    {
      "epoch": 0.12370062370062371,
      "grad_norm": 12.650636672973633,
      "learning_rate": 2.4532224532224534e-06,
      "loss": 2.1807,
      "step": 119
    },
    {
      "epoch": 0.12474012474012475,
      "grad_norm": 16.649805068969727,
      "learning_rate": 2.4740124740124743e-06,
      "loss": 2.1561,
      "step": 120
    },
    {
      "epoch": 0.1257796257796258,
      "grad_norm": 13.328505516052246,
      "learning_rate": 2.4948024948024953e-06,
      "loss": 1.7857,
      "step": 121
    },
    {
      "epoch": 0.12681912681912683,
      "grad_norm": 16.647127151489258,
      "learning_rate": 2.515592515592516e-06,
      "loss": 2.2226,
      "step": 122
    },
    {
      "epoch": 0.12785862785862787,
      "grad_norm": 13.633663177490234,
      "learning_rate": 2.5363825363825367e-06,
      "loss": 1.9883,
      "step": 123
    },
    {
      "epoch": 0.1288981288981289,
      "grad_norm": 11.40990161895752,
      "learning_rate": 2.5571725571725577e-06,
      "loss": 1.9498,
      "step": 124
    },
    {
      "epoch": 0.12993762993762994,
      "grad_norm": 11.874930381774902,
      "learning_rate": 2.577962577962578e-06,
      "loss": 1.7595,
      "step": 125
    },
    {
      "epoch": 0.13097713097713098,
      "grad_norm": 11.909989356994629,
      "learning_rate": 2.598752598752599e-06,
      "loss": 1.7103,
      "step": 126
    },
    {
      "epoch": 0.13201663201663202,
      "grad_norm": 9.262993812561035,
      "learning_rate": 2.61954261954262e-06,
      "loss": 1.8453,
      "step": 127
    },
    {
      "epoch": 0.13305613305613306,
      "grad_norm": 12.807122230529785,
      "learning_rate": 2.6403326403326406e-06,
      "loss": 1.8424,
      "step": 128
    },
    {
      "epoch": 0.1340956340956341,
      "grad_norm": 7.935920715332031,
      "learning_rate": 2.6611226611226616e-06,
      "loss": 1.6972,
      "step": 129
    },
    {
      "epoch": 0.13513513513513514,
      "grad_norm": 11.945698738098145,
      "learning_rate": 2.681912681912682e-06,
      "loss": 1.8897,
      "step": 130
    },
    {
      "epoch": 0.13617463617463618,
      "grad_norm": 12.887810707092285,
      "learning_rate": 2.702702702702703e-06,
      "loss": 1.9529,
      "step": 131
    },
    {
      "epoch": 0.13721413721413722,
      "grad_norm": 10.941514015197754,
      "learning_rate": 2.723492723492724e-06,
      "loss": 1.8406,
      "step": 132
    },
    {
      "epoch": 0.13825363825363826,
      "grad_norm": 10.435567855834961,
      "learning_rate": 2.7442827442827445e-06,
      "loss": 1.5383,
      "step": 133
    },
    {
      "epoch": 0.1392931392931393,
      "grad_norm": 9.996288299560547,
      "learning_rate": 2.7650727650727654e-06,
      "loss": 1.8487,
      "step": 134
    },
    {
      "epoch": 0.14033264033264034,
      "grad_norm": 10.20373249053955,
      "learning_rate": 2.7858627858627864e-06,
      "loss": 1.6566,
      "step": 135
    },
    {
      "epoch": 0.14137214137214138,
      "grad_norm": 13.138648986816406,
      "learning_rate": 2.806652806652807e-06,
      "loss": 1.6186,
      "step": 136
    },
    {
      "epoch": 0.14241164241164242,
      "grad_norm": 11.178861618041992,
      "learning_rate": 2.827442827442828e-06,
      "loss": 1.8976,
      "step": 137
    },
    {
      "epoch": 0.14345114345114346,
      "grad_norm": 7.92245626449585,
      "learning_rate": 2.8482328482328488e-06,
      "loss": 1.5972,
      "step": 138
    },
    {
      "epoch": 0.1444906444906445,
      "grad_norm": 7.310725212097168,
      "learning_rate": 2.8690228690228693e-06,
      "loss": 1.6067,
      "step": 139
    },
    {
      "epoch": 0.14553014553014554,
      "grad_norm": 11.522294998168945,
      "learning_rate": 2.8898128898128902e-06,
      "loss": 1.8074,
      "step": 140
    },
    {
      "epoch": 0.14656964656964658,
      "grad_norm": 11.914559364318848,
      "learning_rate": 2.910602910602911e-06,
      "loss": 1.9732,
      "step": 141
    },
    {
      "epoch": 0.14760914760914762,
      "grad_norm": 12.07341194152832,
      "learning_rate": 2.9313929313929317e-06,
      "loss": 1.7112,
      "step": 142
    },
    {
      "epoch": 0.14864864864864866,
      "grad_norm": 10.485044479370117,
      "learning_rate": 2.9521829521829526e-06,
      "loss": 1.844,
      "step": 143
    },
    {
      "epoch": 0.1496881496881497,
      "grad_norm": 7.178315162658691,
      "learning_rate": 2.9729729729729736e-06,
      "loss": 1.7565,
      "step": 144
    },
    {
      "epoch": 0.15072765072765074,
      "grad_norm": 9.397273063659668,
      "learning_rate": 2.993762993762994e-06,
      "loss": 1.8364,
      "step": 145
    },
    {
      "epoch": 0.15176715176715178,
      "grad_norm": 10.049165725708008,
      "learning_rate": 3.014553014553015e-06,
      "loss": 1.8335,
      "step": 146
    },
    {
      "epoch": 0.15280665280665282,
      "grad_norm": 7.683914661407471,
      "learning_rate": 3.0353430353430356e-06,
      "loss": 1.73,
      "step": 147
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 8.30390739440918,
      "learning_rate": 3.0561330561330565e-06,
      "loss": 1.843,
      "step": 148
    },
    {
      "epoch": 0.1548856548856549,
      "grad_norm": 8.528340339660645,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 1.9296,
      "step": 149
    },
    {
      "epoch": 0.15592515592515593,
      "grad_norm": 11.0097017288208,
      "learning_rate": 3.097713097713098e-06,
      "loss": 2.0037,
      "step": 150
    },
    {
      "epoch": 0.15696465696465697,
      "grad_norm": 7.787356376647949,
      "learning_rate": 3.118503118503119e-06,
      "loss": 1.6986,
      "step": 151
    },
    {
      "epoch": 0.158004158004158,
      "grad_norm": 15.426385879516602,
      "learning_rate": 3.13929313929314e-06,
      "loss": 1.8589,
      "step": 152
    },
    {
      "epoch": 0.15904365904365905,
      "grad_norm": 9.391960144042969,
      "learning_rate": 3.1600831600831604e-06,
      "loss": 1.6876,
      "step": 153
    },
    {
      "epoch": 0.1600831600831601,
      "grad_norm": 12.089409828186035,
      "learning_rate": 3.1808731808731813e-06,
      "loss": 1.4551,
      "step": 154
    },
    {
      "epoch": 0.16112266112266113,
      "grad_norm": 7.866195201873779,
      "learning_rate": 3.2016632016632022e-06,
      "loss": 1.8804,
      "step": 155
    },
    {
      "epoch": 0.16216216216216217,
      "grad_norm": 9.8470458984375,
      "learning_rate": 3.2224532224532228e-06,
      "loss": 1.6656,
      "step": 156
    },
    {
      "epoch": 0.1632016632016632,
      "grad_norm": 7.687832355499268,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 1.4208,
      "step": 157
    },
    {
      "epoch": 0.16424116424116425,
      "grad_norm": 12.187580108642578,
      "learning_rate": 3.2640332640332646e-06,
      "loss": 1.8139,
      "step": 158
    },
    {
      "epoch": 0.1652806652806653,
      "grad_norm": 7.861310958862305,
      "learning_rate": 3.284823284823285e-06,
      "loss": 1.6931,
      "step": 159
    },
    {
      "epoch": 0.16632016632016633,
      "grad_norm": 9.621845245361328,
      "learning_rate": 3.305613305613306e-06,
      "loss": 1.6181,
      "step": 160
    },
    {
      "epoch": 0.16735966735966737,
      "grad_norm": 9.091207504272461,
      "learning_rate": 3.326403326403327e-06,
      "loss": 1.6692,
      "step": 161
    },
    {
      "epoch": 0.1683991683991684,
      "grad_norm": 25.73870849609375,
      "learning_rate": 3.3471933471933476e-06,
      "loss": 1.9534,
      "step": 162
    },
    {
      "epoch": 0.16943866943866945,
      "grad_norm": 17.834957122802734,
      "learning_rate": 3.3679833679833685e-06,
      "loss": 1.8596,
      "step": 163
    },
    {
      "epoch": 0.1704781704781705,
      "grad_norm": 9.130378723144531,
      "learning_rate": 3.388773388773389e-06,
      "loss": 1.6516,
      "step": 164
    },
    {
      "epoch": 0.17151767151767153,
      "grad_norm": 11.076170921325684,
      "learning_rate": 3.40956340956341e-06,
      "loss": 2.0877,
      "step": 165
    },
    {
      "epoch": 0.17255717255717257,
      "grad_norm": 8.953692436218262,
      "learning_rate": 3.430353430353431e-06,
      "loss": 2.0112,
      "step": 166
    },
    {
      "epoch": 0.1735966735966736,
      "grad_norm": 9.42551326751709,
      "learning_rate": 3.4511434511434514e-06,
      "loss": 1.8069,
      "step": 167
    },
    {
      "epoch": 0.17463617463617465,
      "grad_norm": 7.4458794593811035,
      "learning_rate": 3.4719334719334724e-06,
      "loss": 1.5709,
      "step": 168
    },
    {
      "epoch": 0.17567567567567569,
      "grad_norm": 8.580041885375977,
      "learning_rate": 3.4927234927234933e-06,
      "loss": 1.912,
      "step": 169
    },
    {
      "epoch": 0.17671517671517672,
      "grad_norm": 9.437491416931152,
      "learning_rate": 3.513513513513514e-06,
      "loss": 1.6681,
      "step": 170
    },
    {
      "epoch": 0.17775467775467776,
      "grad_norm": 11.995328903198242,
      "learning_rate": 3.5343035343035348e-06,
      "loss": 1.9467,
      "step": 171
    },
    {
      "epoch": 0.1787941787941788,
      "grad_norm": 7.549195766448975,
      "learning_rate": 3.5550935550935557e-06,
      "loss": 1.7225,
      "step": 172
    },
    {
      "epoch": 0.17983367983367984,
      "grad_norm": 9.546483039855957,
      "learning_rate": 3.5758835758835762e-06,
      "loss": 1.8062,
      "step": 173
    },
    {
      "epoch": 0.18087318087318088,
      "grad_norm": 11.021419525146484,
      "learning_rate": 3.596673596673597e-06,
      "loss": 1.858,
      "step": 174
    },
    {
      "epoch": 0.18191268191268192,
      "grad_norm": 17.999780654907227,
      "learning_rate": 3.617463617463618e-06,
      "loss": 1.723,
      "step": 175
    },
    {
      "epoch": 0.18295218295218296,
      "grad_norm": 12.470070838928223,
      "learning_rate": 3.6382536382536386e-06,
      "loss": 1.6866,
      "step": 176
    },
    {
      "epoch": 0.183991683991684,
      "grad_norm": 12.866351127624512,
      "learning_rate": 3.6590436590436596e-06,
      "loss": 1.6734,
      "step": 177
    },
    {
      "epoch": 0.18503118503118504,
      "grad_norm": 7.818451404571533,
      "learning_rate": 3.6798336798336805e-06,
      "loss": 1.6018,
      "step": 178
    },
    {
      "epoch": 0.18607068607068608,
      "grad_norm": 10.40690803527832,
      "learning_rate": 3.700623700623701e-06,
      "loss": 1.6845,
      "step": 179
    },
    {
      "epoch": 0.18711018711018712,
      "grad_norm": 9.794620513916016,
      "learning_rate": 3.721413721413722e-06,
      "loss": 1.8355,
      "step": 180
    },
    {
      "epoch": 0.18814968814968816,
      "grad_norm": 8.024982452392578,
      "learning_rate": 3.7422037422037425e-06,
      "loss": 1.7645,
      "step": 181
    },
    {
      "epoch": 0.1891891891891892,
      "grad_norm": 7.918184280395508,
      "learning_rate": 3.7629937629937634e-06,
      "loss": 1.5322,
      "step": 182
    },
    {
      "epoch": 0.19022869022869024,
      "grad_norm": 8.833913803100586,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 1.6939,
      "step": 183
    },
    {
      "epoch": 0.19126819126819128,
      "grad_norm": 9.3923978805542,
      "learning_rate": 3.804573804573805e-06,
      "loss": 1.453,
      "step": 184
    },
    {
      "epoch": 0.19230769230769232,
      "grad_norm": 10.067290306091309,
      "learning_rate": 3.825363825363826e-06,
      "loss": 1.759,
      "step": 185
    },
    {
      "epoch": 0.19334719334719336,
      "grad_norm": 6.426559925079346,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1.627,
      "step": 186
    },
    {
      "epoch": 0.1943866943866944,
      "grad_norm": 8.969478607177734,
      "learning_rate": 3.866943866943867e-06,
      "loss": 1.6313,
      "step": 187
    },
    {
      "epoch": 0.19542619542619544,
      "grad_norm": 10.542095184326172,
      "learning_rate": 3.887733887733889e-06,
      "loss": 1.6337,
      "step": 188
    },
    {
      "epoch": 0.19646569646569648,
      "grad_norm": 8.737053871154785,
      "learning_rate": 3.908523908523909e-06,
      "loss": 1.5676,
      "step": 189
    },
    {
      "epoch": 0.19750519750519752,
      "grad_norm": 10.72559928894043,
      "learning_rate": 3.92931392931393e-06,
      "loss": 1.6693,
      "step": 190
    },
    {
      "epoch": 0.19854469854469856,
      "grad_norm": 8.796841621398926,
      "learning_rate": 3.95010395010395e-06,
      "loss": 1.6421,
      "step": 191
    },
    {
      "epoch": 0.1995841995841996,
      "grad_norm": 25.390539169311523,
      "learning_rate": 3.970893970893972e-06,
      "loss": 1.5609,
      "step": 192
    },
    {
      "epoch": 0.20062370062370063,
      "grad_norm": 9.860316276550293,
      "learning_rate": 3.991683991683992e-06,
      "loss": 1.8664,
      "step": 193
    },
    {
      "epoch": 0.20166320166320167,
      "grad_norm": 8.280574798583984,
      "learning_rate": 4.012474012474013e-06,
      "loss": 1.5831,
      "step": 194
    },
    {
      "epoch": 0.20270270270270271,
      "grad_norm": 7.029718399047852,
      "learning_rate": 4.033264033264034e-06,
      "loss": 1.6864,
      "step": 195
    },
    {
      "epoch": 0.20374220374220375,
      "grad_norm": 8.269538879394531,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 1.5362,
      "step": 196
    },
    {
      "epoch": 0.2047817047817048,
      "grad_norm": 7.9130940437316895,
      "learning_rate": 4.074844074844075e-06,
      "loss": 1.7531,
      "step": 197
    },
    {
      "epoch": 0.20582120582120583,
      "grad_norm": 9.139872550964355,
      "learning_rate": 4.095634095634096e-06,
      "loss": 1.5782,
      "step": 198
    },
    {
      "epoch": 0.20686070686070687,
      "grad_norm": 8.888442039489746,
      "learning_rate": 4.116424116424117e-06,
      "loss": 1.531,
      "step": 199
    },
    {
      "epoch": 0.2079002079002079,
      "grad_norm": 9.606545448303223,
      "learning_rate": 4.1372141372141374e-06,
      "loss": 1.9533,
      "step": 200
    },
    {
      "epoch": 0.20893970893970895,
      "grad_norm": 7.717611789703369,
      "learning_rate": 4.158004158004159e-06,
      "loss": 1.6154,
      "step": 201
    },
    {
      "epoch": 0.20997920997921,
      "grad_norm": 10.033646583557129,
      "learning_rate": 4.178794178794179e-06,
      "loss": 1.5977,
      "step": 202
    },
    {
      "epoch": 0.21101871101871103,
      "grad_norm": 8.350449562072754,
      "learning_rate": 4.1995841995842e-06,
      "loss": 1.6015,
      "step": 203
    },
    {
      "epoch": 0.21205821205821207,
      "grad_norm": 11.233882904052734,
      "learning_rate": 4.220374220374221e-06,
      "loss": 1.9587,
      "step": 204
    },
    {
      "epoch": 0.2130977130977131,
      "grad_norm": 8.711499214172363,
      "learning_rate": 4.241164241164242e-06,
      "loss": 1.7817,
      "step": 205
    },
    {
      "epoch": 0.21413721413721415,
      "grad_norm": 12.727667808532715,
      "learning_rate": 4.261954261954262e-06,
      "loss": 1.9023,
      "step": 206
    },
    {
      "epoch": 0.2151767151767152,
      "grad_norm": 6.942087650299072,
      "learning_rate": 4.282744282744284e-06,
      "loss": 1.5327,
      "step": 207
    },
    {
      "epoch": 0.21621621621621623,
      "grad_norm": 8.110414505004883,
      "learning_rate": 4.303534303534304e-06,
      "loss": 1.6922,
      "step": 208
    },
    {
      "epoch": 0.21725571725571727,
      "grad_norm": 8.410806655883789,
      "learning_rate": 4.324324324324325e-06,
      "loss": 1.6059,
      "step": 209
    },
    {
      "epoch": 0.2182952182952183,
      "grad_norm": 10.630267143249512,
      "learning_rate": 4.345114345114346e-06,
      "loss": 1.5802,
      "step": 210
    },
    {
      "epoch": 0.21933471933471935,
      "grad_norm": 11.829726219177246,
      "learning_rate": 4.3659043659043665e-06,
      "loss": 1.856,
      "step": 211
    },
    {
      "epoch": 0.2203742203742204,
      "grad_norm": 8.306093215942383,
      "learning_rate": 4.386694386694387e-06,
      "loss": 1.494,
      "step": 212
    },
    {
      "epoch": 0.22141372141372143,
      "grad_norm": 9.279668807983398,
      "learning_rate": 4.4074844074844084e-06,
      "loss": 1.5234,
      "step": 213
    },
    {
      "epoch": 0.22245322245322247,
      "grad_norm": 18.21111297607422,
      "learning_rate": 4.428274428274429e-06,
      "loss": 2.0073,
      "step": 214
    },
    {
      "epoch": 0.2234927234927235,
      "grad_norm": 14.216645240783691,
      "learning_rate": 4.4490644490644495e-06,
      "loss": 1.334,
      "step": 215
    },
    {
      "epoch": 0.22453222453222454,
      "grad_norm": 8.353694915771484,
      "learning_rate": 4.469854469854471e-06,
      "loss": 1.7931,
      "step": 216
    },
    {
      "epoch": 0.22557172557172558,
      "grad_norm": 12.420600891113281,
      "learning_rate": 4.490644490644491e-06,
      "loss": 1.8834,
      "step": 217
    },
    {
      "epoch": 0.22661122661122662,
      "grad_norm": 7.537139892578125,
      "learning_rate": 4.511434511434512e-06,
      "loss": 1.5505,
      "step": 218
    },
    {
      "epoch": 0.22765072765072766,
      "grad_norm": 8.061205863952637,
      "learning_rate": 4.532224532224533e-06,
      "loss": 1.7889,
      "step": 219
    },
    {
      "epoch": 0.2286902286902287,
      "grad_norm": 7.974506378173828,
      "learning_rate": 4.553014553014554e-06,
      "loss": 1.5875,
      "step": 220
    },
    {
      "epoch": 0.22972972972972974,
      "grad_norm": 8.587332725524902,
      "learning_rate": 4.573804573804574e-06,
      "loss": 1.1923,
      "step": 221
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 7.924431800842285,
      "learning_rate": 4.594594594594596e-06,
      "loss": 1.1502,
      "step": 222
    },
    {
      "epoch": 0.23180873180873182,
      "grad_norm": 11.359457015991211,
      "learning_rate": 4.615384615384616e-06,
      "loss": 1.6597,
      "step": 223
    },
    {
      "epoch": 0.23284823284823286,
      "grad_norm": 11.588187217712402,
      "learning_rate": 4.636174636174637e-06,
      "loss": 1.9984,
      "step": 224
    },
    {
      "epoch": 0.2338877338877339,
      "grad_norm": 9.252629280090332,
      "learning_rate": 4.656964656964657e-06,
      "loss": 1.8034,
      "step": 225
    },
    {
      "epoch": 0.23492723492723494,
      "grad_norm": 8.870009422302246,
      "learning_rate": 4.6777546777546786e-06,
      "loss": 1.7183,
      "step": 226
    },
    {
      "epoch": 0.23596673596673598,
      "grad_norm": 6.966862678527832,
      "learning_rate": 4.698544698544699e-06,
      "loss": 1.6209,
      "step": 227
    },
    {
      "epoch": 0.23700623700623702,
      "grad_norm": 6.796176433563232,
      "learning_rate": 4.71933471933472e-06,
      "loss": 1.9011,
      "step": 228
    },
    {
      "epoch": 0.23804573804573806,
      "grad_norm": 8.598974227905273,
      "learning_rate": 4.740124740124741e-06,
      "loss": 1.6843,
      "step": 229
    },
    {
      "epoch": 0.2390852390852391,
      "grad_norm": 8.327461242675781,
      "learning_rate": 4.7609147609147615e-06,
      "loss": 1.737,
      "step": 230
    },
    {
      "epoch": 0.24012474012474014,
      "grad_norm": 9.38319206237793,
      "learning_rate": 4.781704781704782e-06,
      "loss": 1.6075,
      "step": 231
    },
    {
      "epoch": 0.24116424116424118,
      "grad_norm": 5.437087059020996,
      "learning_rate": 4.802494802494803e-06,
      "loss": 1.1498,
      "step": 232
    },
    {
      "epoch": 0.24220374220374222,
      "grad_norm": 8.21915054321289,
      "learning_rate": 4.823284823284824e-06,
      "loss": 1.6603,
      "step": 233
    },
    {
      "epoch": 0.24324324324324326,
      "grad_norm": 8.151537895202637,
      "learning_rate": 4.844074844074844e-06,
      "loss": 1.4857,
      "step": 234
    },
    {
      "epoch": 0.2442827442827443,
      "grad_norm": 7.525641918182373,
      "learning_rate": 4.864864864864866e-06,
      "loss": 1.5105,
      "step": 235
    },
    {
      "epoch": 0.24532224532224534,
      "grad_norm": 7.916234970092773,
      "learning_rate": 4.885654885654886e-06,
      "loss": 1.5264,
      "step": 236
    },
    {
      "epoch": 0.24636174636174638,
      "grad_norm": 9.898544311523438,
      "learning_rate": 4.906444906444907e-06,
      "loss": 1.782,
      "step": 237
    },
    {
      "epoch": 0.24740124740124741,
      "grad_norm": 9.978106498718262,
      "learning_rate": 4.927234927234928e-06,
      "loss": 1.5498,
      "step": 238
    },
    {
      "epoch": 0.24844074844074845,
      "grad_norm": 5.927715301513672,
      "learning_rate": 4.948024948024949e-06,
      "loss": 1.1964,
      "step": 239
    },
    {
      "epoch": 0.2494802494802495,
      "grad_norm": 8.933344841003418,
      "learning_rate": 4.968814968814969e-06,
      "loss": 1.5373,
      "step": 240
    },
    {
      "epoch": 0.2505197505197505,
      "grad_norm": 18.588769912719727,
      "learning_rate": 4.9896049896049906e-06,
      "loss": 1.8564,
      "step": 241
    },
    {
      "epoch": 0.2515592515592516,
      "grad_norm": 8.533500671386719,
      "learning_rate": 5.01039501039501e-06,
      "loss": 1.7192,
      "step": 242
    },
    {
      "epoch": 0.2525987525987526,
      "grad_norm": 9.05661392211914,
      "learning_rate": 5.031185031185032e-06,
      "loss": 1.7023,
      "step": 243
    },
    {
      "epoch": 0.25363825363825365,
      "grad_norm": 7.37769079208374,
      "learning_rate": 5.051975051975052e-06,
      "loss": 1.5155,
      "step": 244
    },
    {
      "epoch": 0.25467775467775466,
      "grad_norm": 8.445049285888672,
      "learning_rate": 5.0727650727650735e-06,
      "loss": 1.4876,
      "step": 245
    },
    {
      "epoch": 0.25571725571725573,
      "grad_norm": 10.22825813293457,
      "learning_rate": 5.093555093555094e-06,
      "loss": 1.658,
      "step": 246
    },
    {
      "epoch": 0.25675675675675674,
      "grad_norm": 8.475323677062988,
      "learning_rate": 5.114345114345115e-06,
      "loss": 1.9086,
      "step": 247
    },
    {
      "epoch": 0.2577962577962578,
      "grad_norm": 9.246431350708008,
      "learning_rate": 5.135135135135135e-06,
      "loss": 1.4162,
      "step": 248
    },
    {
      "epoch": 0.2588357588357588,
      "grad_norm": 11.256653785705566,
      "learning_rate": 5.155925155925156e-06,
      "loss": 1.6352,
      "step": 249
    },
    {
      "epoch": 0.2598752598752599,
      "grad_norm": 8.728947639465332,
      "learning_rate": 5.176715176715177e-06,
      "loss": 1.4578,
      "step": 250
    },
    {
      "epoch": 0.2609147609147609,
      "grad_norm": 10.811807632446289,
      "learning_rate": 5.197505197505198e-06,
      "loss": 1.5422,
      "step": 251
    },
    {
      "epoch": 0.26195426195426197,
      "grad_norm": 8.760584831237793,
      "learning_rate": 5.218295218295219e-06,
      "loss": 1.6168,
      "step": 252
    },
    {
      "epoch": 0.262993762993763,
      "grad_norm": 11.621368408203125,
      "learning_rate": 5.23908523908524e-06,
      "loss": 1.7856,
      "step": 253
    },
    {
      "epoch": 0.26403326403326405,
      "grad_norm": 9.020323753356934,
      "learning_rate": 5.25987525987526e-06,
      "loss": 1.2882,
      "step": 254
    },
    {
      "epoch": 0.26507276507276506,
      "grad_norm": 9.391657829284668,
      "learning_rate": 5.280665280665281e-06,
      "loss": 1.7343,
      "step": 255
    },
    {
      "epoch": 0.2661122661122661,
      "grad_norm": 9.435623168945312,
      "learning_rate": 5.301455301455302e-06,
      "loss": 1.4503,
      "step": 256
    },
    {
      "epoch": 0.26715176715176714,
      "grad_norm": 9.689167022705078,
      "learning_rate": 5.322245322245323e-06,
      "loss": 1.8933,
      "step": 257
    },
    {
      "epoch": 0.2681912681912682,
      "grad_norm": 10.120408058166504,
      "learning_rate": 5.343035343035344e-06,
      "loss": 1.6382,
      "step": 258
    },
    {
      "epoch": 0.2692307692307692,
      "grad_norm": 8.142711639404297,
      "learning_rate": 5.363825363825364e-06,
      "loss": 1.5524,
      "step": 259
    },
    {
      "epoch": 0.2702702702702703,
      "grad_norm": 7.626992702484131,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1.1425,
      "step": 260
    },
    {
      "epoch": 0.2713097713097713,
      "grad_norm": 11.374751091003418,
      "learning_rate": 5.405405405405406e-06,
      "loss": 1.36,
      "step": 261
    },
    {
      "epoch": 0.27234927234927236,
      "grad_norm": 8.555777549743652,
      "learning_rate": 5.4261954261954265e-06,
      "loss": 1.4073,
      "step": 262
    },
    {
      "epoch": 0.2733887733887734,
      "grad_norm": 8.723259925842285,
      "learning_rate": 5.446985446985448e-06,
      "loss": 1.2645,
      "step": 263
    },
    {
      "epoch": 0.27442827442827444,
      "grad_norm": 12.614055633544922,
      "learning_rate": 5.467775467775468e-06,
      "loss": 1.5528,
      "step": 264
    },
    {
      "epoch": 0.27546777546777546,
      "grad_norm": 13.636921882629395,
      "learning_rate": 5.488565488565489e-06,
      "loss": 1.841,
      "step": 265
    },
    {
      "epoch": 0.2765072765072765,
      "grad_norm": 11.47486686706543,
      "learning_rate": 5.5093555093555095e-06,
      "loss": 1.7953,
      "step": 266
    },
    {
      "epoch": 0.27754677754677753,
      "grad_norm": 9.173731803894043,
      "learning_rate": 5.530145530145531e-06,
      "loss": 1.2862,
      "step": 267
    },
    {
      "epoch": 0.2785862785862786,
      "grad_norm": 12.41637897491455,
      "learning_rate": 5.550935550935551e-06,
      "loss": 1.5515,
      "step": 268
    },
    {
      "epoch": 0.2796257796257796,
      "grad_norm": 11.649749755859375,
      "learning_rate": 5.571725571725573e-06,
      "loss": 1.8494,
      "step": 269
    },
    {
      "epoch": 0.2806652806652807,
      "grad_norm": 12.394052505493164,
      "learning_rate": 5.592515592515592e-06,
      "loss": 1.5484,
      "step": 270
    },
    {
      "epoch": 0.2817047817047817,
      "grad_norm": 9.108813285827637,
      "learning_rate": 5.613305613305614e-06,
      "loss": 1.348,
      "step": 271
    },
    {
      "epoch": 0.28274428274428276,
      "grad_norm": 10.887696266174316,
      "learning_rate": 5.634095634095634e-06,
      "loss": 1.3352,
      "step": 272
    },
    {
      "epoch": 0.28378378378378377,
      "grad_norm": 13.775224685668945,
      "learning_rate": 5.654885654885656e-06,
      "loss": 1.3866,
      "step": 273
    },
    {
      "epoch": 0.28482328482328484,
      "grad_norm": 12.727477073669434,
      "learning_rate": 5.675675675675676e-06,
      "loss": 1.1752,
      "step": 274
    },
    {
      "epoch": 0.28586278586278585,
      "grad_norm": 14.530426979064941,
      "learning_rate": 5.6964656964656975e-06,
      "loss": 1.6287,
      "step": 275
    },
    {
      "epoch": 0.2869022869022869,
      "grad_norm": 12.435869216918945,
      "learning_rate": 5.717255717255717e-06,
      "loss": 1.9149,
      "step": 276
    },
    {
      "epoch": 0.28794178794178793,
      "grad_norm": 9.951990127563477,
      "learning_rate": 5.7380457380457386e-06,
      "loss": 1.3319,
      "step": 277
    },
    {
      "epoch": 0.288981288981289,
      "grad_norm": 9.536114692687988,
      "learning_rate": 5.758835758835759e-06,
      "loss": 1.2908,
      "step": 278
    },
    {
      "epoch": 0.29002079002079,
      "grad_norm": 10.469128608703613,
      "learning_rate": 5.7796257796257805e-06,
      "loss": 1.3363,
      "step": 279
    },
    {
      "epoch": 0.2910602910602911,
      "grad_norm": 9.150140762329102,
      "learning_rate": 5.800415800415801e-06,
      "loss": 1.376,
      "step": 280
    },
    {
      "epoch": 0.2920997920997921,
      "grad_norm": 9.672096252441406,
      "learning_rate": 5.821205821205822e-06,
      "loss": 1.4166,
      "step": 281
    },
    {
      "epoch": 0.29313929313929316,
      "grad_norm": 7.898129940032959,
      "learning_rate": 5.841995841995842e-06,
      "loss": 1.1823,
      "step": 282
    },
    {
      "epoch": 0.29417879417879417,
      "grad_norm": 12.452738761901855,
      "learning_rate": 5.862785862785863e-06,
      "loss": 1.5592,
      "step": 283
    },
    {
      "epoch": 0.29521829521829523,
      "grad_norm": 8.472002029418945,
      "learning_rate": 5.883575883575884e-06,
      "loss": 1.2292,
      "step": 284
    },
    {
      "epoch": 0.29625779625779625,
      "grad_norm": 10.24071979522705,
      "learning_rate": 5.904365904365905e-06,
      "loss": 1.4567,
      "step": 285
    },
    {
      "epoch": 0.2972972972972973,
      "grad_norm": 13.903212547302246,
      "learning_rate": 5.925155925155926e-06,
      "loss": 1.7304,
      "step": 286
    },
    {
      "epoch": 0.2983367983367983,
      "grad_norm": 13.026294708251953,
      "learning_rate": 5.945945945945947e-06,
      "loss": 1.3533,
      "step": 287
    },
    {
      "epoch": 0.2993762993762994,
      "grad_norm": 8.777679443359375,
      "learning_rate": 5.966735966735967e-06,
      "loss": 1.3465,
      "step": 288
    },
    {
      "epoch": 0.3004158004158004,
      "grad_norm": 8.598836898803711,
      "learning_rate": 5.987525987525988e-06,
      "loss": 1.5693,
      "step": 289
    },
    {
      "epoch": 0.30145530145530147,
      "grad_norm": 12.184210777282715,
      "learning_rate": 6.008316008316009e-06,
      "loss": 1.3762,
      "step": 290
    },
    {
      "epoch": 0.3024948024948025,
      "grad_norm": 10.718639373779297,
      "learning_rate": 6.02910602910603e-06,
      "loss": 1.7274,
      "step": 291
    },
    {
      "epoch": 0.30353430353430355,
      "grad_norm": 7.593832492828369,
      "learning_rate": 6.049896049896051e-06,
      "loss": 1.625,
      "step": 292
    },
    {
      "epoch": 0.30457380457380456,
      "grad_norm": 11.308578491210938,
      "learning_rate": 6.070686070686071e-06,
      "loss": 1.3614,
      "step": 293
    },
    {
      "epoch": 0.30561330561330563,
      "grad_norm": 8.035221099853516,
      "learning_rate": 6.091476091476092e-06,
      "loss": 1.2422,
      "step": 294
    },
    {
      "epoch": 0.30665280665280664,
      "grad_norm": 8.732400894165039,
      "learning_rate": 6.112266112266113e-06,
      "loss": 1.6587,
      "step": 295
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 8.046781539916992,
      "learning_rate": 6.1330561330561335e-06,
      "loss": 1.4965,
      "step": 296
    },
    {
      "epoch": 0.3087318087318087,
      "grad_norm": 7.799867153167725,
      "learning_rate": 6.153846153846155e-06,
      "loss": 1.2193,
      "step": 297
    },
    {
      "epoch": 0.3097713097713098,
      "grad_norm": 10.253771781921387,
      "learning_rate": 6.1746361746361745e-06,
      "loss": 1.7422,
      "step": 298
    },
    {
      "epoch": 0.3108108108108108,
      "grad_norm": 11.052000999450684,
      "learning_rate": 6.195426195426196e-06,
      "loss": 1.1786,
      "step": 299
    },
    {
      "epoch": 0.31185031185031187,
      "grad_norm": 13.860453605651855,
      "learning_rate": 6.2162162162162164e-06,
      "loss": 1.4553,
      "step": 300
    },
    {
      "epoch": 0.3128898128898129,
      "grad_norm": 10.874279022216797,
      "learning_rate": 6.237006237006238e-06,
      "loss": 1.2194,
      "step": 301
    },
    {
      "epoch": 0.31392931392931395,
      "grad_norm": 8.688791275024414,
      "learning_rate": 6.257796257796258e-06,
      "loss": 1.3588,
      "step": 302
    },
    {
      "epoch": 0.31496881496881496,
      "grad_norm": 12.43751335144043,
      "learning_rate": 6.27858627858628e-06,
      "loss": 1.1144,
      "step": 303
    },
    {
      "epoch": 0.316008316008316,
      "grad_norm": 11.306876182556152,
      "learning_rate": 6.299376299376299e-06,
      "loss": 1.4284,
      "step": 304
    },
    {
      "epoch": 0.31704781704781704,
      "grad_norm": 11.36539077758789,
      "learning_rate": 6.320166320166321e-06,
      "loss": 1.4414,
      "step": 305
    },
    {
      "epoch": 0.3180873180873181,
      "grad_norm": 11.967350006103516,
      "learning_rate": 6.340956340956341e-06,
      "loss": 1.43,
      "step": 306
    },
    {
      "epoch": 0.3191268191268191,
      "grad_norm": 8.277191162109375,
      "learning_rate": 6.361746361746363e-06,
      "loss": 1.0795,
      "step": 307
    },
    {
      "epoch": 0.3201663201663202,
      "grad_norm": 13.161870002746582,
      "learning_rate": 6.382536382536383e-06,
      "loss": 1.4126,
      "step": 308
    },
    {
      "epoch": 0.3212058212058212,
      "grad_norm": 8.901982307434082,
      "learning_rate": 6.4033264033264045e-06,
      "loss": 1.0459,
      "step": 309
    },
    {
      "epoch": 0.32224532224532226,
      "grad_norm": 9.05133056640625,
      "learning_rate": 6.424116424116424e-06,
      "loss": 0.8949,
      "step": 310
    },
    {
      "epoch": 0.3232848232848233,
      "grad_norm": 11.73928165435791,
      "learning_rate": 6.4449064449064455e-06,
      "loss": 1.1428,
      "step": 311
    },
    {
      "epoch": 0.32432432432432434,
      "grad_norm": 11.88116455078125,
      "learning_rate": 6.465696465696466e-06,
      "loss": 1.2491,
      "step": 312
    },
    {
      "epoch": 0.32536382536382535,
      "grad_norm": 8.605231285095215,
      "learning_rate": 6.486486486486487e-06,
      "loss": 1.0301,
      "step": 313
    },
    {
      "epoch": 0.3264033264033264,
      "grad_norm": 12.880175590515137,
      "learning_rate": 6.507276507276508e-06,
      "loss": 1.3983,
      "step": 314
    },
    {
      "epoch": 0.32744282744282743,
      "grad_norm": 8.701872825622559,
      "learning_rate": 6.528066528066529e-06,
      "loss": 1.5113,
      "step": 315
    },
    {
      "epoch": 0.3284823284823285,
      "grad_norm": 8.418421745300293,
      "learning_rate": 6.548856548856549e-06,
      "loss": 1.1952,
      "step": 316
    },
    {
      "epoch": 0.3295218295218295,
      "grad_norm": 12.062923431396484,
      "learning_rate": 6.56964656964657e-06,
      "loss": 1.3726,
      "step": 317
    },
    {
      "epoch": 0.3305613305613306,
      "grad_norm": 10.043243408203125,
      "learning_rate": 6.590436590436591e-06,
      "loss": 1.295,
      "step": 318
    },
    {
      "epoch": 0.3316008316008316,
      "grad_norm": 11.718816757202148,
      "learning_rate": 6.611226611226612e-06,
      "loss": 0.8623,
      "step": 319
    },
    {
      "epoch": 0.33264033264033266,
      "grad_norm": 11.69416332244873,
      "learning_rate": 6.632016632016633e-06,
      "loss": 0.9903,
      "step": 320
    },
    {
      "epoch": 0.33367983367983367,
      "grad_norm": 12.75631046295166,
      "learning_rate": 6.652806652806654e-06,
      "loss": 0.8313,
      "step": 321
    },
    {
      "epoch": 0.33471933471933474,
      "grad_norm": 8.642829895019531,
      "learning_rate": 6.673596673596674e-06,
      "loss": 0.8052,
      "step": 322
    },
    {
      "epoch": 0.33575883575883575,
      "grad_norm": 7.5102152824401855,
      "learning_rate": 6.694386694386695e-06,
      "loss": 0.812,
      "step": 323
    },
    {
      "epoch": 0.3367983367983368,
      "grad_norm": 15.602274894714355,
      "learning_rate": 6.715176715176716e-06,
      "loss": 0.8896,
      "step": 324
    },
    {
      "epoch": 0.33783783783783783,
      "grad_norm": 19.65711784362793,
      "learning_rate": 6.735966735966737e-06,
      "loss": 1.3914,
      "step": 325
    },
    {
      "epoch": 0.3388773388773389,
      "grad_norm": 8.210640907287598,
      "learning_rate": 6.7567567567567575e-06,
      "loss": 1.0448,
      "step": 326
    },
    {
      "epoch": 0.3399168399168399,
      "grad_norm": 15.147570610046387,
      "learning_rate": 6.777546777546778e-06,
      "loss": 1.1706,
      "step": 327
    },
    {
      "epoch": 0.340956340956341,
      "grad_norm": 29.13080406188965,
      "learning_rate": 6.7983367983367986e-06,
      "loss": 1.3041,
      "step": 328
    },
    {
      "epoch": 0.341995841995842,
      "grad_norm": 16.381385803222656,
      "learning_rate": 6.81912681912682e-06,
      "loss": 1.0187,
      "step": 329
    },
    {
      "epoch": 0.34303534303534305,
      "grad_norm": 13.570858001708984,
      "learning_rate": 6.8399168399168405e-06,
      "loss": 1.3256,
      "step": 330
    },
    {
      "epoch": 0.34407484407484407,
      "grad_norm": 15.603581428527832,
      "learning_rate": 6.860706860706862e-06,
      "loss": 1.0822,
      "step": 331
    },
    {
      "epoch": 0.34511434511434513,
      "grad_norm": 21.508617401123047,
      "learning_rate": 6.8814968814968815e-06,
      "loss": 1.5841,
      "step": 332
    },
    {
      "epoch": 0.34615384615384615,
      "grad_norm": 9.773581504821777,
      "learning_rate": 6.902286902286903e-06,
      "loss": 0.7203,
      "step": 333
    },
    {
      "epoch": 0.3471933471933472,
      "grad_norm": 17.405637741088867,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.9002,
      "step": 334
    },
    {
      "epoch": 0.3482328482328482,
      "grad_norm": 16.036029815673828,
      "learning_rate": 6.943866943866945e-06,
      "loss": 1.1254,
      "step": 335
    },
    {
      "epoch": 0.3492723492723493,
      "grad_norm": 8.478259086608887,
      "learning_rate": 6.964656964656965e-06,
      "loss": 0.7698,
      "step": 336
    },
    {
      "epoch": 0.3503118503118503,
      "grad_norm": 11.398589134216309,
      "learning_rate": 6.985446985446987e-06,
      "loss": 0.8371,
      "step": 337
    },
    {
      "epoch": 0.35135135135135137,
      "grad_norm": 10.781030654907227,
      "learning_rate": 7.006237006237006e-06,
      "loss": 1.2204,
      "step": 338
    },
    {
      "epoch": 0.3523908523908524,
      "grad_norm": 13.479138374328613,
      "learning_rate": 7.027027027027028e-06,
      "loss": 1.3343,
      "step": 339
    },
    {
      "epoch": 0.35343035343035345,
      "grad_norm": 10.525655746459961,
      "learning_rate": 7.047817047817048e-06,
      "loss": 1.4545,
      "step": 340
    },
    {
      "epoch": 0.35446985446985446,
      "grad_norm": 11.65915298461914,
      "learning_rate": 7.0686070686070696e-06,
      "loss": 0.7949,
      "step": 341
    },
    {
      "epoch": 0.35550935550935553,
      "grad_norm": 20.972801208496094,
      "learning_rate": 7.08939708939709e-06,
      "loss": 1.2358,
      "step": 342
    },
    {
      "epoch": 0.35654885654885654,
      "grad_norm": 12.074674606323242,
      "learning_rate": 7.1101871101871114e-06,
      "loss": 0.8076,
      "step": 343
    },
    {
      "epoch": 0.3575883575883576,
      "grad_norm": 12.374785423278809,
      "learning_rate": 7.130977130977131e-06,
      "loss": 1.2175,
      "step": 344
    },
    {
      "epoch": 0.3586278586278586,
      "grad_norm": 12.02391242980957,
      "learning_rate": 7.1517671517671525e-06,
      "loss": 0.6736,
      "step": 345
    },
    {
      "epoch": 0.3596673596673597,
      "grad_norm": 14.542448043823242,
      "learning_rate": 7.172557172557173e-06,
      "loss": 0.9352,
      "step": 346
    },
    {
      "epoch": 0.3607068607068607,
      "grad_norm": 25.29374122619629,
      "learning_rate": 7.193347193347194e-06,
      "loss": 1.2332,
      "step": 347
    },
    {
      "epoch": 0.36174636174636177,
      "grad_norm": 10.751561164855957,
      "learning_rate": 7.214137214137215e-06,
      "loss": 1.0363,
      "step": 348
    },
    {
      "epoch": 0.3627858627858628,
      "grad_norm": 6.808762550354004,
      "learning_rate": 7.234927234927236e-06,
      "loss": 0.5515,
      "step": 349
    },
    {
      "epoch": 0.36382536382536385,
      "grad_norm": 15.856203079223633,
      "learning_rate": 7.255717255717256e-06,
      "loss": 0.9221,
      "step": 350
    },
    {
      "epoch": 0.36486486486486486,
      "grad_norm": 9.114971160888672,
      "learning_rate": 7.276507276507277e-06,
      "loss": 0.8534,
      "step": 351
    },
    {
      "epoch": 0.3659043659043659,
      "grad_norm": 14.44951057434082,
      "learning_rate": 7.297297297297298e-06,
      "loss": 1.0257,
      "step": 352
    },
    {
      "epoch": 0.36694386694386694,
      "grad_norm": 10.981499671936035,
      "learning_rate": 7.318087318087319e-06,
      "loss": 1.1816,
      "step": 353
    },
    {
      "epoch": 0.367983367983368,
      "grad_norm": 11.531734466552734,
      "learning_rate": 7.33887733887734e-06,
      "loss": 1.1302,
      "step": 354
    },
    {
      "epoch": 0.369022869022869,
      "grad_norm": 14.686858177185059,
      "learning_rate": 7.359667359667361e-06,
      "loss": 1.2317,
      "step": 355
    },
    {
      "epoch": 0.3700623700623701,
      "grad_norm": 6.359785079956055,
      "learning_rate": 7.380457380457381e-06,
      "loss": 0.6505,
      "step": 356
    },
    {
      "epoch": 0.3711018711018711,
      "grad_norm": 9.21711540222168,
      "learning_rate": 7.401247401247402e-06,
      "loss": 0.902,
      "step": 357
    },
    {
      "epoch": 0.37214137214137216,
      "grad_norm": 7.829357147216797,
      "learning_rate": 7.422037422037423e-06,
      "loss": 0.7133,
      "step": 358
    },
    {
      "epoch": 0.3731808731808732,
      "grad_norm": 14.824617385864258,
      "learning_rate": 7.442827442827444e-06,
      "loss": 1.5177,
      "step": 359
    },
    {
      "epoch": 0.37422037422037424,
      "grad_norm": 13.023225784301758,
      "learning_rate": 7.4636174636174645e-06,
      "loss": 1.1179,
      "step": 360
    },
    {
      "epoch": 0.37525987525987525,
      "grad_norm": 11.518128395080566,
      "learning_rate": 7.484407484407485e-06,
      "loss": 0.9857,
      "step": 361
    },
    {
      "epoch": 0.3762993762993763,
      "grad_norm": 9.43225383758545,
      "learning_rate": 7.5051975051975055e-06,
      "loss": 0.9234,
      "step": 362
    },
    {
      "epoch": 0.37733887733887733,
      "grad_norm": 10.689541816711426,
      "learning_rate": 7.525987525987527e-06,
      "loss": 0.9689,
      "step": 363
    },
    {
      "epoch": 0.3783783783783784,
      "grad_norm": 12.340954780578613,
      "learning_rate": 7.546777546777547e-06,
      "loss": 0.887,
      "step": 364
    },
    {
      "epoch": 0.3794178794178794,
      "grad_norm": 7.20005464553833,
      "learning_rate": 7.567567567567569e-06,
      "loss": 0.5674,
      "step": 365
    },
    {
      "epoch": 0.3804573804573805,
      "grad_norm": 10.14612865447998,
      "learning_rate": 7.5883575883575885e-06,
      "loss": 0.7206,
      "step": 366
    },
    {
      "epoch": 0.3814968814968815,
      "grad_norm": 19.117307662963867,
      "learning_rate": 7.60914760914761e-06,
      "loss": 1.5382,
      "step": 367
    },
    {
      "epoch": 0.38253638253638256,
      "grad_norm": 13.663631439208984,
      "learning_rate": 7.629937629937631e-06,
      "loss": 1.1351,
      "step": 368
    },
    {
      "epoch": 0.38357588357588357,
      "grad_norm": 10.310964584350586,
      "learning_rate": 7.650727650727653e-06,
      "loss": 1.2692,
      "step": 369
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 7.663285255432129,
      "learning_rate": 7.671517671517672e-06,
      "loss": 0.4554,
      "step": 370
    },
    {
      "epoch": 0.38565488565488565,
      "grad_norm": 10.936558723449707,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.8166,
      "step": 371
    },
    {
      "epoch": 0.3866943866943867,
      "grad_norm": 6.133451461791992,
      "learning_rate": 7.713097713097713e-06,
      "loss": 0.6886,
      "step": 372
    },
    {
      "epoch": 0.3877338877338877,
      "grad_norm": 9.430103302001953,
      "learning_rate": 7.733887733887735e-06,
      "loss": 1.0831,
      "step": 373
    },
    {
      "epoch": 0.3887733887733888,
      "grad_norm": 6.969508647918701,
      "learning_rate": 7.754677754677756e-06,
      "loss": 0.65,
      "step": 374
    },
    {
      "epoch": 0.3898128898128898,
      "grad_norm": 8.69268798828125,
      "learning_rate": 7.775467775467777e-06,
      "loss": 1.1201,
      "step": 375
    },
    {
      "epoch": 0.3908523908523909,
      "grad_norm": 6.384047031402588,
      "learning_rate": 7.796257796257797e-06,
      "loss": 0.5161,
      "step": 376
    },
    {
      "epoch": 0.3918918918918919,
      "grad_norm": 8.49499797821045,
      "learning_rate": 7.817047817047818e-06,
      "loss": 0.6475,
      "step": 377
    },
    {
      "epoch": 0.39293139293139295,
      "grad_norm": 7.770305633544922,
      "learning_rate": 7.837837837837838e-06,
      "loss": 0.597,
      "step": 378
    },
    {
      "epoch": 0.39397089397089397,
      "grad_norm": 10.950555801391602,
      "learning_rate": 7.85862785862786e-06,
      "loss": 1.07,
      "step": 379
    },
    {
      "epoch": 0.39501039501039503,
      "grad_norm": 9.929052352905273,
      "learning_rate": 7.879417879417879e-06,
      "loss": 0.9038,
      "step": 380
    },
    {
      "epoch": 0.39604989604989604,
      "grad_norm": 10.960211753845215,
      "learning_rate": 7.9002079002079e-06,
      "loss": 1.0922,
      "step": 381
    },
    {
      "epoch": 0.3970893970893971,
      "grad_norm": 13.324213027954102,
      "learning_rate": 7.920997920997922e-06,
      "loss": 1.3425,
      "step": 382
    },
    {
      "epoch": 0.3981288981288981,
      "grad_norm": 11.701374053955078,
      "learning_rate": 7.941787941787943e-06,
      "loss": 0.9576,
      "step": 383
    },
    {
      "epoch": 0.3991683991683992,
      "grad_norm": 11.246499061584473,
      "learning_rate": 7.962577962577963e-06,
      "loss": 1.0604,
      "step": 384
    },
    {
      "epoch": 0.4002079002079002,
      "grad_norm": 6.898358345031738,
      "learning_rate": 7.983367983367984e-06,
      "loss": 0.4263,
      "step": 385
    },
    {
      "epoch": 0.40124740124740127,
      "grad_norm": 7.588034152984619,
      "learning_rate": 8.004158004158004e-06,
      "loss": 0.6515,
      "step": 386
    },
    {
      "epoch": 0.4022869022869023,
      "grad_norm": 8.791478157043457,
      "learning_rate": 8.024948024948025e-06,
      "loss": 0.5638,
      "step": 387
    },
    {
      "epoch": 0.40332640332640335,
      "grad_norm": 12.265979766845703,
      "learning_rate": 8.045738045738047e-06,
      "loss": 0.9777,
      "step": 388
    },
    {
      "epoch": 0.40436590436590436,
      "grad_norm": 12.479912757873535,
      "learning_rate": 8.066528066528068e-06,
      "loss": 1.2196,
      "step": 389
    },
    {
      "epoch": 0.40540540540540543,
      "grad_norm": 10.367191314697266,
      "learning_rate": 8.087318087318088e-06,
      "loss": 1.1416,
      "step": 390
    },
    {
      "epoch": 0.40644490644490644,
      "grad_norm": 6.253833770751953,
      "learning_rate": 8.108108108108109e-06,
      "loss": 0.4802,
      "step": 391
    },
    {
      "epoch": 0.4074844074844075,
      "grad_norm": 14.616800308227539,
      "learning_rate": 8.128898128898129e-06,
      "loss": 1.2278,
      "step": 392
    },
    {
      "epoch": 0.4085239085239085,
      "grad_norm": 12.594634056091309,
      "learning_rate": 8.14968814968815e-06,
      "loss": 1.0137,
      "step": 393
    },
    {
      "epoch": 0.4095634095634096,
      "grad_norm": 11.438323020935059,
      "learning_rate": 8.170478170478171e-06,
      "loss": 1.019,
      "step": 394
    },
    {
      "epoch": 0.4106029106029106,
      "grad_norm": 13.236550331115723,
      "learning_rate": 8.191268191268193e-06,
      "loss": 0.7806,
      "step": 395
    },
    {
      "epoch": 0.41164241164241167,
      "grad_norm": 4.43021297454834,
      "learning_rate": 8.212058212058212e-06,
      "loss": 0.2711,
      "step": 396
    },
    {
      "epoch": 0.4126819126819127,
      "grad_norm": 7.386447429656982,
      "learning_rate": 8.232848232848234e-06,
      "loss": 0.5482,
      "step": 397
    },
    {
      "epoch": 0.41372141372141374,
      "grad_norm": 12.167444229125977,
      "learning_rate": 8.253638253638254e-06,
      "loss": 1.1778,
      "step": 398
    },
    {
      "epoch": 0.41476091476091476,
      "grad_norm": 12.794252395629883,
      "learning_rate": 8.274428274428275e-06,
      "loss": 0.8015,
      "step": 399
    },
    {
      "epoch": 0.4158004158004158,
      "grad_norm": 10.782106399536133,
      "learning_rate": 8.295218295218296e-06,
      "loss": 1.0166,
      "step": 400
    },
    {
      "epoch": 0.41683991683991684,
      "grad_norm": 9.652565956115723,
      "learning_rate": 8.316008316008318e-06,
      "loss": 0.6135,
      "step": 401
    },
    {
      "epoch": 0.4178794178794179,
      "grad_norm": 10.046876907348633,
      "learning_rate": 8.336798336798337e-06,
      "loss": 0.6397,
      "step": 402
    },
    {
      "epoch": 0.4189189189189189,
      "grad_norm": 3.3654232025146484,
      "learning_rate": 8.357588357588359e-06,
      "loss": 0.1272,
      "step": 403
    },
    {
      "epoch": 0.41995841995842,
      "grad_norm": 8.895442008972168,
      "learning_rate": 8.378378378378378e-06,
      "loss": 0.3683,
      "step": 404
    },
    {
      "epoch": 0.420997920997921,
      "grad_norm": 8.768726348876953,
      "learning_rate": 8.3991683991684e-06,
      "loss": 0.4138,
      "step": 405
    },
    {
      "epoch": 0.42203742203742206,
      "grad_norm": 8.897137641906738,
      "learning_rate": 8.419958419958421e-06,
      "loss": 0.5625,
      "step": 406
    },
    {
      "epoch": 0.4230769230769231,
      "grad_norm": 9.527857780456543,
      "learning_rate": 8.440748440748442e-06,
      "loss": 0.5531,
      "step": 407
    },
    {
      "epoch": 0.42411642411642414,
      "grad_norm": 21.51705551147461,
      "learning_rate": 8.461538461538462e-06,
      "loss": 1.2941,
      "step": 408
    },
    {
      "epoch": 0.42515592515592515,
      "grad_norm": 22.675201416015625,
      "learning_rate": 8.482328482328483e-06,
      "loss": 1.3033,
      "step": 409
    },
    {
      "epoch": 0.4261954261954262,
      "grad_norm": 7.007496356964111,
      "learning_rate": 8.503118503118503e-06,
      "loss": 0.2945,
      "step": 410
    },
    {
      "epoch": 0.42723492723492723,
      "grad_norm": 9.356853485107422,
      "learning_rate": 8.523908523908525e-06,
      "loss": 0.4295,
      "step": 411
    },
    {
      "epoch": 0.4282744282744283,
      "grad_norm": 12.781023025512695,
      "learning_rate": 8.544698544698546e-06,
      "loss": 0.8006,
      "step": 412
    },
    {
      "epoch": 0.4293139293139293,
      "grad_norm": 15.798759460449219,
      "learning_rate": 8.565488565488567e-06,
      "loss": 0.871,
      "step": 413
    },
    {
      "epoch": 0.4303534303534304,
      "grad_norm": 12.773930549621582,
      "learning_rate": 8.586278586278587e-06,
      "loss": 0.5821,
      "step": 414
    },
    {
      "epoch": 0.4313929313929314,
      "grad_norm": 10.541467666625977,
      "learning_rate": 8.607068607068608e-06,
      "loss": 0.6077,
      "step": 415
    },
    {
      "epoch": 0.43243243243243246,
      "grad_norm": 9.462098121643066,
      "learning_rate": 8.627858627858628e-06,
      "loss": 0.3352,
      "step": 416
    },
    {
      "epoch": 0.43347193347193347,
      "grad_norm": 10.985498428344727,
      "learning_rate": 8.64864864864865e-06,
      "loss": 0.4576,
      "step": 417
    },
    {
      "epoch": 0.43451143451143454,
      "grad_norm": 15.112650871276855,
      "learning_rate": 8.66943866943867e-06,
      "loss": 0.8339,
      "step": 418
    },
    {
      "epoch": 0.43555093555093555,
      "grad_norm": 13.53800106048584,
      "learning_rate": 8.690228690228692e-06,
      "loss": 0.7046,
      "step": 419
    },
    {
      "epoch": 0.4365904365904366,
      "grad_norm": 15.45318603515625,
      "learning_rate": 8.711018711018712e-06,
      "loss": 0.6397,
      "step": 420
    },
    {
      "epoch": 0.4376299376299376,
      "grad_norm": 1.3829221725463867,
      "learning_rate": 8.731808731808733e-06,
      "loss": 0.0236,
      "step": 421
    },
    {
      "epoch": 0.4386694386694387,
      "grad_norm": 5.065361976623535,
      "learning_rate": 8.752598752598753e-06,
      "loss": 0.161,
      "step": 422
    },
    {
      "epoch": 0.4397089397089397,
      "grad_norm": 6.4144368171691895,
      "learning_rate": 8.773388773388774e-06,
      "loss": 0.1571,
      "step": 423
    },
    {
      "epoch": 0.4407484407484408,
      "grad_norm": 11.337523460388184,
      "learning_rate": 8.794178794178795e-06,
      "loss": 0.4098,
      "step": 424
    },
    {
      "epoch": 0.4417879417879418,
      "grad_norm": 10.023104667663574,
      "learning_rate": 8.814968814968817e-06,
      "loss": 0.6923,
      "step": 425
    },
    {
      "epoch": 0.44282744282744285,
      "grad_norm": 3.946394920349121,
      "learning_rate": 8.835758835758837e-06,
      "loss": 0.0945,
      "step": 426
    },
    {
      "epoch": 0.44386694386694386,
      "grad_norm": 17.10584259033203,
      "learning_rate": 8.856548856548858e-06,
      "loss": 0.8002,
      "step": 427
    },
    {
      "epoch": 0.44490644490644493,
      "grad_norm": 11.162398338317871,
      "learning_rate": 8.877338877338878e-06,
      "loss": 0.3048,
      "step": 428
    },
    {
      "epoch": 0.44594594594594594,
      "grad_norm": 9.242036819458008,
      "learning_rate": 8.898128898128899e-06,
      "loss": 0.1708,
      "step": 429
    },
    {
      "epoch": 0.446985446985447,
      "grad_norm": 1.562695860862732,
      "learning_rate": 8.91891891891892e-06,
      "loss": 0.0244,
      "step": 430
    },
    {
      "epoch": 0.448024948024948,
      "grad_norm": 9.712590217590332,
      "learning_rate": 8.939708939708942e-06,
      "loss": 0.1978,
      "step": 431
    },
    {
      "epoch": 0.4490644490644491,
      "grad_norm": 12.932109832763672,
      "learning_rate": 8.960498960498961e-06,
      "loss": 0.3598,
      "step": 432
    },
    {
      "epoch": 0.4501039501039501,
      "grad_norm": 7.306901454925537,
      "learning_rate": 8.981288981288983e-06,
      "loss": 0.1072,
      "step": 433
    },
    {
      "epoch": 0.45114345114345117,
      "grad_norm": 18.84465980529785,
      "learning_rate": 9.002079002079002e-06,
      "loss": 0.3701,
      "step": 434
    },
    {
      "epoch": 0.4521829521829522,
      "grad_norm": 13.76640796661377,
      "learning_rate": 9.022869022869024e-06,
      "loss": 0.3655,
      "step": 435
    },
    {
      "epoch": 0.45322245322245325,
      "grad_norm": 11.665372848510742,
      "learning_rate": 9.043659043659045e-06,
      "loss": 0.5584,
      "step": 436
    },
    {
      "epoch": 0.45426195426195426,
      "grad_norm": 15.73073959350586,
      "learning_rate": 9.064449064449066e-06,
      "loss": 0.7605,
      "step": 437
    },
    {
      "epoch": 0.4553014553014553,
      "grad_norm": 11.39462661743164,
      "learning_rate": 9.085239085239086e-06,
      "loss": 0.1828,
      "step": 438
    },
    {
      "epoch": 0.45634095634095634,
      "grad_norm": 13.786553382873535,
      "learning_rate": 9.106029106029107e-06,
      "loss": 0.4993,
      "step": 439
    },
    {
      "epoch": 0.4573804573804574,
      "grad_norm": 19.499631881713867,
      "learning_rate": 9.126819126819127e-06,
      "loss": 1.396,
      "step": 440
    },
    {
      "epoch": 0.4584199584199584,
      "grad_norm": 20.219745635986328,
      "learning_rate": 9.147609147609149e-06,
      "loss": 0.7873,
      "step": 441
    },
    {
      "epoch": 0.4594594594594595,
      "grad_norm": 18.218852996826172,
      "learning_rate": 9.16839916839917e-06,
      "loss": 0.4743,
      "step": 442
    },
    {
      "epoch": 0.4604989604989605,
      "grad_norm": 19.086950302124023,
      "learning_rate": 9.189189189189191e-06,
      "loss": 0.3544,
      "step": 443
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 18.145187377929688,
      "learning_rate": 9.209979209979211e-06,
      "loss": 0.6076,
      "step": 444
    },
    {
      "epoch": 0.4625779625779626,
      "grad_norm": 7.627259731292725,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.1382,
      "step": 445
    },
    {
      "epoch": 0.46361746361746364,
      "grad_norm": 6.8412017822265625,
      "learning_rate": 9.251559251559252e-06,
      "loss": 0.1242,
      "step": 446
    },
    {
      "epoch": 0.46465696465696466,
      "grad_norm": 16.168813705444336,
      "learning_rate": 9.272349272349273e-06,
      "loss": 0.8019,
      "step": 447
    },
    {
      "epoch": 0.4656964656964657,
      "grad_norm": 13.182543754577637,
      "learning_rate": 9.293139293139293e-06,
      "loss": 0.2616,
      "step": 448
    },
    {
      "epoch": 0.46673596673596673,
      "grad_norm": 13.589136123657227,
      "learning_rate": 9.313929313929314e-06,
      "loss": 0.5943,
      "step": 449
    },
    {
      "epoch": 0.4677754677754678,
      "grad_norm": 15.259722709655762,
      "learning_rate": 9.334719334719336e-06,
      "loss": 0.2538,
      "step": 450
    },
    {
      "epoch": 0.4688149688149688,
      "grad_norm": 21.816892623901367,
      "learning_rate": 9.355509355509357e-06,
      "loss": 0.9268,
      "step": 451
    },
    {
      "epoch": 0.4698544698544699,
      "grad_norm": 13.662321090698242,
      "learning_rate": 9.376299376299377e-06,
      "loss": 0.4554,
      "step": 452
    },
    {
      "epoch": 0.4708939708939709,
      "grad_norm": 19.907556533813477,
      "learning_rate": 9.397089397089398e-06,
      "loss": 0.8886,
      "step": 453
    },
    {
      "epoch": 0.47193347193347196,
      "grad_norm": 11.883735656738281,
      "learning_rate": 9.417879417879418e-06,
      "loss": 0.1615,
      "step": 454
    },
    {
      "epoch": 0.47297297297297297,
      "grad_norm": 2.11840558052063,
      "learning_rate": 9.43866943866944e-06,
      "loss": 0.0334,
      "step": 455
    },
    {
      "epoch": 0.47401247401247404,
      "grad_norm": 23.978952407836914,
      "learning_rate": 9.45945945945946e-06,
      "loss": 0.9978,
      "step": 456
    },
    {
      "epoch": 0.47505197505197505,
      "grad_norm": 17.58618927001953,
      "learning_rate": 9.480249480249482e-06,
      "loss": 0.5197,
      "step": 457
    },
    {
      "epoch": 0.4760914760914761,
      "grad_norm": 7.399749279022217,
      "learning_rate": 9.501039501039502e-06,
      "loss": 0.1385,
      "step": 458
    },
    {
      "epoch": 0.47713097713097713,
      "grad_norm": 2.330472469329834,
      "learning_rate": 9.521829521829523e-06,
      "loss": 0.0326,
      "step": 459
    },
    {
      "epoch": 0.4781704781704782,
      "grad_norm": 8.512271881103516,
      "learning_rate": 9.542619542619543e-06,
      "loss": 0.2566,
      "step": 460
    },
    {
      "epoch": 0.4792099792099792,
      "grad_norm": 15.216105461120605,
      "learning_rate": 9.563409563409564e-06,
      "loss": 0.3799,
      "step": 461
    },
    {
      "epoch": 0.4802494802494803,
      "grad_norm": 27.830585479736328,
      "learning_rate": 9.584199584199585e-06,
      "loss": 0.7251,
      "step": 462
    },
    {
      "epoch": 0.4812889812889813,
      "grad_norm": 11.564122200012207,
      "learning_rate": 9.604989604989607e-06,
      "loss": 0.2673,
      "step": 463
    },
    {
      "epoch": 0.48232848232848236,
      "grad_norm": 19.60334587097168,
      "learning_rate": 9.625779625779626e-06,
      "loss": 0.5008,
      "step": 464
    },
    {
      "epoch": 0.48336798336798337,
      "grad_norm": 6.509657382965088,
      "learning_rate": 9.646569646569648e-06,
      "loss": 0.1444,
      "step": 465
    },
    {
      "epoch": 0.48440748440748443,
      "grad_norm": 11.038702011108398,
      "learning_rate": 9.667359667359667e-06,
      "loss": 0.3259,
      "step": 466
    },
    {
      "epoch": 0.48544698544698545,
      "grad_norm": 13.629114151000977,
      "learning_rate": 9.688149688149689e-06,
      "loss": 0.4144,
      "step": 467
    },
    {
      "epoch": 0.4864864864864865,
      "grad_norm": 3.2071421146392822,
      "learning_rate": 9.70893970893971e-06,
      "loss": 0.0468,
      "step": 468
    },
    {
      "epoch": 0.4875259875259875,
      "grad_norm": 14.734481811523438,
      "learning_rate": 9.729729729729732e-06,
      "loss": 0.4032,
      "step": 469
    },
    {
      "epoch": 0.4885654885654886,
      "grad_norm": 11.064592361450195,
      "learning_rate": 9.750519750519751e-06,
      "loss": 0.2775,
      "step": 470
    },
    {
      "epoch": 0.4896049896049896,
      "grad_norm": 18.813310623168945,
      "learning_rate": 9.771309771309773e-06,
      "loss": 0.7688,
      "step": 471
    },
    {
      "epoch": 0.49064449064449067,
      "grad_norm": 13.356475830078125,
      "learning_rate": 9.792099792099792e-06,
      "loss": 0.2939,
      "step": 472
    },
    {
      "epoch": 0.4916839916839917,
      "grad_norm": 6.419248104095459,
      "learning_rate": 9.812889812889814e-06,
      "loss": 0.1099,
      "step": 473
    },
    {
      "epoch": 0.49272349272349275,
      "grad_norm": 16.916677474975586,
      "learning_rate": 9.833679833679835e-06,
      "loss": 0.5852,
      "step": 474
    },
    {
      "epoch": 0.49376299376299376,
      "grad_norm": 16.16600799560547,
      "learning_rate": 9.854469854469856e-06,
      "loss": 0.3559,
      "step": 475
    },
    {
      "epoch": 0.49480249480249483,
      "grad_norm": 18.88541030883789,
      "learning_rate": 9.875259875259876e-06,
      "loss": 0.475,
      "step": 476
    },
    {
      "epoch": 0.49584199584199584,
      "grad_norm": 7.591237545013428,
      "learning_rate": 9.896049896049897e-06,
      "loss": 0.1283,
      "step": 477
    },
    {
      "epoch": 0.4968814968814969,
      "grad_norm": 23.891620635986328,
      "learning_rate": 9.916839916839917e-06,
      "loss": 0.7828,
      "step": 478
    },
    {
      "epoch": 0.4979209979209979,
      "grad_norm": 7.804129600524902,
      "learning_rate": 9.937629937629938e-06,
      "loss": 0.0996,
      "step": 479
    },
    {
      "epoch": 0.498960498960499,
      "grad_norm": 15.474300384521484,
      "learning_rate": 9.95841995841996e-06,
      "loss": 0.4721,
      "step": 480
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.639959335327148,
      "learning_rate": 9.979209979209981e-06,
      "loss": 0.1883,
      "step": 481
    },
    {
      "epoch": 0.501039501039501,
      "grad_norm": 19.550806045532227,
      "learning_rate": 1e-05,
      "loss": 0.6499,
      "step": 482
    },
    {
      "epoch": 0.502079002079002,
      "grad_norm": 17.1366024017334,
      "learning_rate": 9.997689997689999e-06,
      "loss": 0.6938,
      "step": 483
    },
    {
      "epoch": 0.5031185031185031,
      "grad_norm": 10.473664283752441,
      "learning_rate": 9.995379995379996e-06,
      "loss": 0.3482,
      "step": 484
    },
    {
      "epoch": 0.5041580041580042,
      "grad_norm": 35.813411712646484,
      "learning_rate": 9.993069993069994e-06,
      "loss": 1.2533,
      "step": 485
    },
    {
      "epoch": 0.5051975051975052,
      "grad_norm": 10.875179290771484,
      "learning_rate": 9.990759990759992e-06,
      "loss": 0.4058,
      "step": 486
    },
    {
      "epoch": 0.5062370062370062,
      "grad_norm": 7.9114580154418945,
      "learning_rate": 9.98844998844999e-06,
      "loss": 0.1588,
      "step": 487
    },
    {
      "epoch": 0.5072765072765073,
      "grad_norm": 21.617277145385742,
      "learning_rate": 9.986139986139987e-06,
      "loss": 0.5731,
      "step": 488
    },
    {
      "epoch": 0.5083160083160083,
      "grad_norm": 11.489331245422363,
      "learning_rate": 9.983829983829985e-06,
      "loss": 0.1674,
      "step": 489
    },
    {
      "epoch": 0.5093555093555093,
      "grad_norm": 23.25141716003418,
      "learning_rate": 9.981519981519983e-06,
      "loss": 0.4789,
      "step": 490
    },
    {
      "epoch": 0.5103950103950103,
      "grad_norm": 14.97988224029541,
      "learning_rate": 9.979209979209981e-06,
      "loss": 0.2813,
      "step": 491
    },
    {
      "epoch": 0.5114345114345115,
      "grad_norm": 21.924203872680664,
      "learning_rate": 9.976899976899978e-06,
      "loss": 0.7728,
      "step": 492
    },
    {
      "epoch": 0.5124740124740125,
      "grad_norm": 21.031352996826172,
      "learning_rate": 9.974589974589976e-06,
      "loss": 0.6349,
      "step": 493
    },
    {
      "epoch": 0.5135135135135135,
      "grad_norm": 6.328756332397461,
      "learning_rate": 9.972279972279974e-06,
      "loss": 0.0946,
      "step": 494
    },
    {
      "epoch": 0.5145530145530145,
      "grad_norm": 14.493115425109863,
      "learning_rate": 9.96996996996997e-06,
      "loss": 0.2316,
      "step": 495
    },
    {
      "epoch": 0.5155925155925156,
      "grad_norm": 19.657384872436523,
      "learning_rate": 9.967659967659969e-06,
      "loss": 0.5465,
      "step": 496
    },
    {
      "epoch": 0.5166320166320166,
      "grad_norm": 9.537667274475098,
      "learning_rate": 9.965349965349967e-06,
      "loss": 0.1388,
      "step": 497
    },
    {
      "epoch": 0.5176715176715176,
      "grad_norm": 4.219407081604004,
      "learning_rate": 9.963039963039965e-06,
      "loss": 0.1253,
      "step": 498
    },
    {
      "epoch": 0.5187110187110187,
      "grad_norm": 9.209030151367188,
      "learning_rate": 9.960729960729962e-06,
      "loss": 0.1181,
      "step": 499
    },
    {
      "epoch": 0.5197505197505198,
      "grad_norm": 4.694278717041016,
      "learning_rate": 9.95841995841996e-06,
      "loss": 0.1103,
      "step": 500
    },
    {
      "epoch": 0.5207900207900208,
      "grad_norm": 16.673931121826172,
      "learning_rate": 9.956109956109958e-06,
      "loss": 0.5307,
      "step": 501
    },
    {
      "epoch": 0.5218295218295218,
      "grad_norm": 0.8699054718017578,
      "learning_rate": 9.953799953799954e-06,
      "loss": 0.0086,
      "step": 502
    },
    {
      "epoch": 0.5228690228690228,
      "grad_norm": 7.444122791290283,
      "learning_rate": 9.951489951489953e-06,
      "loss": 0.2038,
      "step": 503
    },
    {
      "epoch": 0.5239085239085239,
      "grad_norm": 20.159868240356445,
      "learning_rate": 9.949179949179951e-06,
      "loss": 0.4744,
      "step": 504
    },
    {
      "epoch": 0.524948024948025,
      "grad_norm": 21.489547729492188,
      "learning_rate": 9.946869946869947e-06,
      "loss": 0.688,
      "step": 505
    },
    {
      "epoch": 0.525987525987526,
      "grad_norm": 31.612947463989258,
      "learning_rate": 9.944559944559946e-06,
      "loss": 1.2621,
      "step": 506
    },
    {
      "epoch": 0.527027027027027,
      "grad_norm": 14.285310745239258,
      "learning_rate": 9.942249942249944e-06,
      "loss": 0.2198,
      "step": 507
    },
    {
      "epoch": 0.5280665280665281,
      "grad_norm": 18.622798919677734,
      "learning_rate": 9.93993993993994e-06,
      "loss": 0.6707,
      "step": 508
    },
    {
      "epoch": 0.5291060291060291,
      "grad_norm": 8.829988479614258,
      "learning_rate": 9.937629937629938e-06,
      "loss": 0.2278,
      "step": 509
    },
    {
      "epoch": 0.5301455301455301,
      "grad_norm": 17.411701202392578,
      "learning_rate": 9.935319935319937e-06,
      "loss": 0.3917,
      "step": 510
    },
    {
      "epoch": 0.5311850311850311,
      "grad_norm": 10.52638053894043,
      "learning_rate": 9.933009933009933e-06,
      "loss": 0.317,
      "step": 511
    },
    {
      "epoch": 0.5322245322245323,
      "grad_norm": 9.510506629943848,
      "learning_rate": 9.930699930699931e-06,
      "loss": 0.1242,
      "step": 512
    },
    {
      "epoch": 0.5332640332640333,
      "grad_norm": 9.235204696655273,
      "learning_rate": 9.92838992838993e-06,
      "loss": 0.1194,
      "step": 513
    },
    {
      "epoch": 0.5343035343035343,
      "grad_norm": 2.6380372047424316,
      "learning_rate": 9.926079926079926e-06,
      "loss": 0.0322,
      "step": 514
    },
    {
      "epoch": 0.5353430353430353,
      "grad_norm": 8.815104484558105,
      "learning_rate": 9.923769923769924e-06,
      "loss": 0.2188,
      "step": 515
    },
    {
      "epoch": 0.5363825363825364,
      "grad_norm": 13.807114601135254,
      "learning_rate": 9.921459921459922e-06,
      "loss": 0.3807,
      "step": 516
    },
    {
      "epoch": 0.5374220374220374,
      "grad_norm": 2.663893222808838,
      "learning_rate": 9.919149919149919e-06,
      "loss": 0.0458,
      "step": 517
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 1.582839846611023,
      "learning_rate": 9.916839916839917e-06,
      "loss": 0.0466,
      "step": 518
    },
    {
      "epoch": 0.5395010395010394,
      "grad_norm": 15.220515251159668,
      "learning_rate": 9.914529914529915e-06,
      "loss": 0.3472,
      "step": 519
    },
    {
      "epoch": 0.5405405405405406,
      "grad_norm": 14.371718406677246,
      "learning_rate": 9.912219912219912e-06,
      "loss": 0.2532,
      "step": 520
    },
    {
      "epoch": 0.5415800415800416,
      "grad_norm": 7.996894359588623,
      "learning_rate": 9.90990990990991e-06,
      "loss": 0.1664,
      "step": 521
    },
    {
      "epoch": 0.5426195426195426,
      "grad_norm": 20.90623664855957,
      "learning_rate": 9.907599907599908e-06,
      "loss": 0.8354,
      "step": 522
    },
    {
      "epoch": 0.5436590436590436,
      "grad_norm": 13.043829917907715,
      "learning_rate": 9.905289905289906e-06,
      "loss": 0.1403,
      "step": 523
    },
    {
      "epoch": 0.5446985446985447,
      "grad_norm": 6.000689506530762,
      "learning_rate": 9.902979902979903e-06,
      "loss": 0.0659,
      "step": 524
    },
    {
      "epoch": 0.5457380457380457,
      "grad_norm": 19.005416870117188,
      "learning_rate": 9.900669900669901e-06,
      "loss": 0.783,
      "step": 525
    },
    {
      "epoch": 0.5467775467775468,
      "grad_norm": 9.024373054504395,
      "learning_rate": 9.8983598983599e-06,
      "loss": 0.1836,
      "step": 526
    },
    {
      "epoch": 0.5478170478170478,
      "grad_norm": 18.351057052612305,
      "learning_rate": 9.896049896049897e-06,
      "loss": 0.277,
      "step": 527
    },
    {
      "epoch": 0.5488565488565489,
      "grad_norm": 14.205464363098145,
      "learning_rate": 9.893739893739894e-06,
      "loss": 0.3695,
      "step": 528
    },
    {
      "epoch": 0.5498960498960499,
      "grad_norm": 15.938841819763184,
      "learning_rate": 9.891429891429892e-06,
      "loss": 0.5653,
      "step": 529
    },
    {
      "epoch": 0.5509355509355509,
      "grad_norm": 11.979267120361328,
      "learning_rate": 9.88911988911989e-06,
      "loss": 0.2696,
      "step": 530
    },
    {
      "epoch": 0.5519750519750519,
      "grad_norm": 22.4202880859375,
      "learning_rate": 9.886809886809887e-06,
      "loss": 0.899,
      "step": 531
    },
    {
      "epoch": 0.553014553014553,
      "grad_norm": 11.355657577514648,
      "learning_rate": 9.884499884499885e-06,
      "loss": 0.2037,
      "step": 532
    },
    {
      "epoch": 0.5540540540540541,
      "grad_norm": 3.479240655899048,
      "learning_rate": 9.882189882189883e-06,
      "loss": 0.0641,
      "step": 533
    },
    {
      "epoch": 0.5550935550935551,
      "grad_norm": 4.595727920532227,
      "learning_rate": 9.879879879879881e-06,
      "loss": 0.0648,
      "step": 534
    },
    {
      "epoch": 0.5561330561330561,
      "grad_norm": 10.914381980895996,
      "learning_rate": 9.877569877569878e-06,
      "loss": 0.2337,
      "step": 535
    },
    {
      "epoch": 0.5571725571725572,
      "grad_norm": 16.96885108947754,
      "learning_rate": 9.875259875259876e-06,
      "loss": 0.5575,
      "step": 536
    },
    {
      "epoch": 0.5582120582120582,
      "grad_norm": 11.56965160369873,
      "learning_rate": 9.872949872949874e-06,
      "loss": 0.2081,
      "step": 537
    },
    {
      "epoch": 0.5592515592515592,
      "grad_norm": 5.558108806610107,
      "learning_rate": 9.87063987063987e-06,
      "loss": 0.0941,
      "step": 538
    },
    {
      "epoch": 0.5602910602910602,
      "grad_norm": 27.092920303344727,
      "learning_rate": 9.868329868329869e-06,
      "loss": 1.0123,
      "step": 539
    },
    {
      "epoch": 0.5613305613305614,
      "grad_norm": 13.499934196472168,
      "learning_rate": 9.866019866019867e-06,
      "loss": 0.2519,
      "step": 540
    },
    {
      "epoch": 0.5623700623700624,
      "grad_norm": 1.136060118675232,
      "learning_rate": 9.863709863709865e-06,
      "loss": 0.0161,
      "step": 541
    },
    {
      "epoch": 0.5634095634095634,
      "grad_norm": 13.566397666931152,
      "learning_rate": 9.861399861399862e-06,
      "loss": 0.3632,
      "step": 542
    },
    {
      "epoch": 0.5644490644490644,
      "grad_norm": 8.34858512878418,
      "learning_rate": 9.85908985908986e-06,
      "loss": 0.1934,
      "step": 543
    },
    {
      "epoch": 0.5654885654885655,
      "grad_norm": 12.781171798706055,
      "learning_rate": 9.856779856779858e-06,
      "loss": 0.185,
      "step": 544
    },
    {
      "epoch": 0.5665280665280665,
      "grad_norm": 15.231748580932617,
      "learning_rate": 9.854469854469856e-06,
      "loss": 0.4749,
      "step": 545
    },
    {
      "epoch": 0.5675675675675675,
      "grad_norm": 9.513189315795898,
      "learning_rate": 9.852159852159853e-06,
      "loss": 0.2114,
      "step": 546
    },
    {
      "epoch": 0.5686070686070686,
      "grad_norm": 3.4976556301116943,
      "learning_rate": 9.849849849849851e-06,
      "loss": 0.0485,
      "step": 547
    },
    {
      "epoch": 0.5696465696465697,
      "grad_norm": 16.309349060058594,
      "learning_rate": 9.84753984753985e-06,
      "loss": 0.2677,
      "step": 548
    },
    {
      "epoch": 0.5706860706860707,
      "grad_norm": 14.880147933959961,
      "learning_rate": 9.845229845229846e-06,
      "loss": 0.4456,
      "step": 549
    },
    {
      "epoch": 0.5717255717255717,
      "grad_norm": 16.281221389770508,
      "learning_rate": 9.842919842919844e-06,
      "loss": 0.4056,
      "step": 550
    },
    {
      "epoch": 0.5727650727650727,
      "grad_norm": 20.214174270629883,
      "learning_rate": 9.840609840609842e-06,
      "loss": 0.4839,
      "step": 551
    },
    {
      "epoch": 0.5738045738045738,
      "grad_norm": 22.193727493286133,
      "learning_rate": 9.83829983829984e-06,
      "loss": 1.3147,
      "step": 552
    },
    {
      "epoch": 0.5748440748440748,
      "grad_norm": 18.85350227355957,
      "learning_rate": 9.835989835989837e-06,
      "loss": 0.3605,
      "step": 553
    },
    {
      "epoch": 0.5758835758835759,
      "grad_norm": 10.782029151916504,
      "learning_rate": 9.833679833679835e-06,
      "loss": 0.202,
      "step": 554
    },
    {
      "epoch": 0.5769230769230769,
      "grad_norm": 12.252769470214844,
      "learning_rate": 9.831369831369833e-06,
      "loss": 0.2616,
      "step": 555
    },
    {
      "epoch": 0.577962577962578,
      "grad_norm": 15.310741424560547,
      "learning_rate": 9.82905982905983e-06,
      "loss": 0.3184,
      "step": 556
    },
    {
      "epoch": 0.579002079002079,
      "grad_norm": 1.3405869007110596,
      "learning_rate": 9.826749826749828e-06,
      "loss": 0.0113,
      "step": 557
    },
    {
      "epoch": 0.58004158004158,
      "grad_norm": 5.590155124664307,
      "learning_rate": 9.824439824439826e-06,
      "loss": 0.1065,
      "step": 558
    },
    {
      "epoch": 0.581081081081081,
      "grad_norm": 16.842506408691406,
      "learning_rate": 9.822129822129823e-06,
      "loss": 0.3817,
      "step": 559
    },
    {
      "epoch": 0.5821205821205822,
      "grad_norm": 10.099876403808594,
      "learning_rate": 9.81981981981982e-06,
      "loss": 0.2883,
      "step": 560
    },
    {
      "epoch": 0.5831600831600832,
      "grad_norm": 21.479961395263672,
      "learning_rate": 9.817509817509819e-06,
      "loss": 1.5162,
      "step": 561
    },
    {
      "epoch": 0.5841995841995842,
      "grad_norm": 7.661498546600342,
      "learning_rate": 9.815199815199815e-06,
      "loss": 0.1604,
      "step": 562
    },
    {
      "epoch": 0.5852390852390852,
      "grad_norm": 11.383598327636719,
      "learning_rate": 9.812889812889814e-06,
      "loss": 0.2646,
      "step": 563
    },
    {
      "epoch": 0.5862785862785863,
      "grad_norm": 20.18816375732422,
      "learning_rate": 9.810579810579812e-06,
      "loss": 0.1702,
      "step": 564
    },
    {
      "epoch": 0.5873180873180873,
      "grad_norm": 16.200801849365234,
      "learning_rate": 9.808269808269808e-06,
      "loss": 1.7284,
      "step": 565
    },
    {
      "epoch": 0.5883575883575883,
      "grad_norm": 3.830944776535034,
      "learning_rate": 9.805959805959806e-06,
      "loss": 0.0424,
      "step": 566
    },
    {
      "epoch": 0.5893970893970893,
      "grad_norm": 13.652180671691895,
      "learning_rate": 9.803649803649805e-06,
      "loss": 0.3261,
      "step": 567
    },
    {
      "epoch": 0.5904365904365905,
      "grad_norm": 1.004449486732483,
      "learning_rate": 9.801339801339801e-06,
      "loss": 0.0188,
      "step": 568
    },
    {
      "epoch": 0.5914760914760915,
      "grad_norm": 1.3249083757400513,
      "learning_rate": 9.7990297990298e-06,
      "loss": 0.0224,
      "step": 569
    },
    {
      "epoch": 0.5925155925155925,
      "grad_norm": 1.8351414203643799,
      "learning_rate": 9.796719796719798e-06,
      "loss": 0.0229,
      "step": 570
    },
    {
      "epoch": 0.5935550935550935,
      "grad_norm": 3.9487857818603516,
      "learning_rate": 9.794409794409794e-06,
      "loss": 0.2799,
      "step": 571
    },
    {
      "epoch": 0.5945945945945946,
      "grad_norm": 3.0859389305114746,
      "learning_rate": 9.792099792099792e-06,
      "loss": 0.0573,
      "step": 572
    },
    {
      "epoch": 0.5956340956340956,
      "grad_norm": 17.17660903930664,
      "learning_rate": 9.78978978978979e-06,
      "loss": 0.6055,
      "step": 573
    },
    {
      "epoch": 0.5966735966735967,
      "grad_norm": 0.18140429258346558,
      "learning_rate": 9.787479787479787e-06,
      "loss": 0.0015,
      "step": 574
    },
    {
      "epoch": 0.5977130977130977,
      "grad_norm": 29.232139587402344,
      "learning_rate": 9.785169785169785e-06,
      "loss": 1.2985,
      "step": 575
    },
    {
      "epoch": 0.5987525987525988,
      "grad_norm": 19.850553512573242,
      "learning_rate": 9.782859782859783e-06,
      "loss": 0.5502,
      "step": 576
    },
    {
      "epoch": 0.5997920997920998,
      "grad_norm": 11.339086532592773,
      "learning_rate": 9.780549780549782e-06,
      "loss": 0.2392,
      "step": 577
    },
    {
      "epoch": 0.6008316008316008,
      "grad_norm": 27.784828186035156,
      "learning_rate": 9.778239778239778e-06,
      "loss": 1.231,
      "step": 578
    },
    {
      "epoch": 0.6018711018711018,
      "grad_norm": 8.059019088745117,
      "learning_rate": 9.775929775929776e-06,
      "loss": 0.1051,
      "step": 579
    },
    {
      "epoch": 0.6029106029106029,
      "grad_norm": 22.404436111450195,
      "learning_rate": 9.773619773619774e-06,
      "loss": 0.8404,
      "step": 580
    },
    {
      "epoch": 0.603950103950104,
      "grad_norm": 19.912567138671875,
      "learning_rate": 9.771309771309773e-06,
      "loss": 1.4489,
      "step": 581
    },
    {
      "epoch": 0.604989604989605,
      "grad_norm": 15.487265586853027,
      "learning_rate": 9.768999768999769e-06,
      "loss": 0.63,
      "step": 582
    },
    {
      "epoch": 0.606029106029106,
      "grad_norm": 21.618928909301758,
      "learning_rate": 9.766689766689767e-06,
      "loss": 1.554,
      "step": 583
    },
    {
      "epoch": 0.6070686070686071,
      "grad_norm": 13.416857719421387,
      "learning_rate": 9.764379764379765e-06,
      "loss": 0.1677,
      "step": 584
    },
    {
      "epoch": 0.6081081081081081,
      "grad_norm": 2.621568441390991,
      "learning_rate": 9.762069762069762e-06,
      "loss": 0.0388,
      "step": 585
    },
    {
      "epoch": 0.6091476091476091,
      "grad_norm": 9.277111053466797,
      "learning_rate": 9.75975975975976e-06,
      "loss": 0.2088,
      "step": 586
    },
    {
      "epoch": 0.6101871101871101,
      "grad_norm": 4.179928779602051,
      "learning_rate": 9.757449757449758e-06,
      "loss": 0.0586,
      "step": 587
    },
    {
      "epoch": 0.6112266112266113,
      "grad_norm": 6.063162326812744,
      "learning_rate": 9.755139755139757e-06,
      "loss": 0.1165,
      "step": 588
    },
    {
      "epoch": 0.6122661122661123,
      "grad_norm": 10.404489517211914,
      "learning_rate": 9.752829752829753e-06,
      "loss": 0.2205,
      "step": 589
    },
    {
      "epoch": 0.6133056133056133,
      "grad_norm": 9.35464859008789,
      "learning_rate": 9.750519750519751e-06,
      "loss": 0.1712,
      "step": 590
    },
    {
      "epoch": 0.6143451143451143,
      "grad_norm": 11.338526725769043,
      "learning_rate": 9.74820974820975e-06,
      "loss": 0.3342,
      "step": 591
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 10.987810134887695,
      "learning_rate": 9.745899745899746e-06,
      "loss": 0.1355,
      "step": 592
    },
    {
      "epoch": 0.6164241164241164,
      "grad_norm": 18.870922088623047,
      "learning_rate": 9.743589743589744e-06,
      "loss": 0.8299,
      "step": 593
    },
    {
      "epoch": 0.6174636174636174,
      "grad_norm": 18.805932998657227,
      "learning_rate": 9.741279741279742e-06,
      "loss": 0.6295,
      "step": 594
    },
    {
      "epoch": 0.6185031185031185,
      "grad_norm": 32.01840591430664,
      "learning_rate": 9.73896973896974e-06,
      "loss": 2.1948,
      "step": 595
    },
    {
      "epoch": 0.6195426195426196,
      "grad_norm": 16.707277297973633,
      "learning_rate": 9.736659736659737e-06,
      "loss": 0.3787,
      "step": 596
    },
    {
      "epoch": 0.6205821205821206,
      "grad_norm": 7.447889804840088,
      "learning_rate": 9.734349734349735e-06,
      "loss": 0.1948,
      "step": 597
    },
    {
      "epoch": 0.6216216216216216,
      "grad_norm": 18.314401626586914,
      "learning_rate": 9.732039732039733e-06,
      "loss": 1.1807,
      "step": 598
    },
    {
      "epoch": 0.6226611226611226,
      "grad_norm": 2.9496583938598633,
      "learning_rate": 9.729729729729732e-06,
      "loss": 0.0373,
      "step": 599
    },
    {
      "epoch": 0.6237006237006237,
      "grad_norm": 7.875022888183594,
      "learning_rate": 9.727419727419728e-06,
      "loss": 0.1404,
      "step": 600
    },
    {
      "epoch": 0.6247401247401247,
      "grad_norm": 22.426799774169922,
      "learning_rate": 9.725109725109726e-06,
      "loss": 0.5067,
      "step": 601
    },
    {
      "epoch": 0.6257796257796258,
      "grad_norm": 2.4129889011383057,
      "learning_rate": 9.722799722799724e-06,
      "loss": 0.0515,
      "step": 602
    },
    {
      "epoch": 0.6268191268191268,
      "grad_norm": 0.15341880917549133,
      "learning_rate": 9.720489720489721e-06,
      "loss": 0.0028,
      "step": 603
    },
    {
      "epoch": 0.6278586278586279,
      "grad_norm": 3.9162347316741943,
      "learning_rate": 9.718179718179719e-06,
      "loss": 0.0555,
      "step": 604
    },
    {
      "epoch": 0.6288981288981289,
      "grad_norm": 7.17239236831665,
      "learning_rate": 9.715869715869717e-06,
      "loss": 0.1233,
      "step": 605
    },
    {
      "epoch": 0.6299376299376299,
      "grad_norm": 2.595475673675537,
      "learning_rate": 9.713559713559715e-06,
      "loss": 0.0454,
      "step": 606
    },
    {
      "epoch": 0.6309771309771309,
      "grad_norm": 19.02918243408203,
      "learning_rate": 9.711249711249712e-06,
      "loss": 0.4381,
      "step": 607
    },
    {
      "epoch": 0.632016632016632,
      "grad_norm": 12.635756492614746,
      "learning_rate": 9.70893970893971e-06,
      "loss": 0.221,
      "step": 608
    },
    {
      "epoch": 0.6330561330561331,
      "grad_norm": 19.13248634338379,
      "learning_rate": 9.706629706629708e-06,
      "loss": 0.5459,
      "step": 609
    },
    {
      "epoch": 0.6340956340956341,
      "grad_norm": 19.797473907470703,
      "learning_rate": 9.704319704319705e-06,
      "loss": 1.0616,
      "step": 610
    },
    {
      "epoch": 0.6351351351351351,
      "grad_norm": 8.363163948059082,
      "learning_rate": 9.702009702009703e-06,
      "loss": 0.3124,
      "step": 611
    },
    {
      "epoch": 0.6361746361746362,
      "grad_norm": 14.620701789855957,
      "learning_rate": 9.699699699699701e-06,
      "loss": 0.9523,
      "step": 612
    },
    {
      "epoch": 0.6372141372141372,
      "grad_norm": 0.11327558010816574,
      "learning_rate": 9.6973896973897e-06,
      "loss": 0.0025,
      "step": 613
    },
    {
      "epoch": 0.6382536382536382,
      "grad_norm": 7.28572416305542,
      "learning_rate": 9.695079695079696e-06,
      "loss": 0.1897,
      "step": 614
    },
    {
      "epoch": 0.6392931392931392,
      "grad_norm": 14.20917797088623,
      "learning_rate": 9.692769692769694e-06,
      "loss": 0.7104,
      "step": 615
    },
    {
      "epoch": 0.6403326403326404,
      "grad_norm": 3.038438320159912,
      "learning_rate": 9.690459690459692e-06,
      "loss": 0.0491,
      "step": 616
    },
    {
      "epoch": 0.6413721413721414,
      "grad_norm": 2.8274102210998535,
      "learning_rate": 9.688149688149689e-06,
      "loss": 0.0379,
      "step": 617
    },
    {
      "epoch": 0.6424116424116424,
      "grad_norm": 8.627979278564453,
      "learning_rate": 9.685839685839687e-06,
      "loss": 0.153,
      "step": 618
    },
    {
      "epoch": 0.6434511434511434,
      "grad_norm": 11.217168807983398,
      "learning_rate": 9.683529683529685e-06,
      "loss": 0.2238,
      "step": 619
    },
    {
      "epoch": 0.6444906444906445,
      "grad_norm": 8.23984432220459,
      "learning_rate": 9.681219681219682e-06,
      "loss": 0.1156,
      "step": 620
    },
    {
      "epoch": 0.6455301455301455,
      "grad_norm": 1.6440142393112183,
      "learning_rate": 9.67890967890968e-06,
      "loss": 0.021,
      "step": 621
    },
    {
      "epoch": 0.6465696465696466,
      "grad_norm": 1.043070673942566,
      "learning_rate": 9.676599676599678e-06,
      "loss": 0.0115,
      "step": 622
    },
    {
      "epoch": 0.6476091476091476,
      "grad_norm": 10.603458404541016,
      "learning_rate": 9.674289674289675e-06,
      "loss": 0.2493,
      "step": 623
    },
    {
      "epoch": 0.6486486486486487,
      "grad_norm": 11.63265609741211,
      "learning_rate": 9.671979671979673e-06,
      "loss": 0.3605,
      "step": 624
    },
    {
      "epoch": 0.6496881496881497,
      "grad_norm": 21.488149642944336,
      "learning_rate": 9.669669669669671e-06,
      "loss": 1.1281,
      "step": 625
    },
    {
      "epoch": 0.6507276507276507,
      "grad_norm": 5.428545951843262,
      "learning_rate": 9.667359667359667e-06,
      "loss": 0.0537,
      "step": 626
    },
    {
      "epoch": 0.6517671517671517,
      "grad_norm": 13.269102096557617,
      "learning_rate": 9.665049665049666e-06,
      "loss": 0.336,
      "step": 627
    },
    {
      "epoch": 0.6528066528066528,
      "grad_norm": 6.222421646118164,
      "learning_rate": 9.662739662739664e-06,
      "loss": 0.0742,
      "step": 628
    },
    {
      "epoch": 0.6538461538461539,
      "grad_norm": 2.2896013259887695,
      "learning_rate": 9.66042966042966e-06,
      "loss": 0.0411,
      "step": 629
    },
    {
      "epoch": 0.6548856548856549,
      "grad_norm": 14.19056224822998,
      "learning_rate": 9.658119658119659e-06,
      "loss": 0.4373,
      "step": 630
    },
    {
      "epoch": 0.6559251559251559,
      "grad_norm": 14.150191307067871,
      "learning_rate": 9.655809655809657e-06,
      "loss": 0.5048,
      "step": 631
    },
    {
      "epoch": 0.656964656964657,
      "grad_norm": 8.00671672821045,
      "learning_rate": 9.653499653499653e-06,
      "loss": 0.1181,
      "step": 632
    },
    {
      "epoch": 0.658004158004158,
      "grad_norm": 14.54723072052002,
      "learning_rate": 9.651189651189651e-06,
      "loss": 0.9277,
      "step": 633
    },
    {
      "epoch": 0.659043659043659,
      "grad_norm": 0.5207761526107788,
      "learning_rate": 9.64887964887965e-06,
      "loss": 0.0086,
      "step": 634
    },
    {
      "epoch": 0.66008316008316,
      "grad_norm": 15.758940696716309,
      "learning_rate": 9.646569646569648e-06,
      "loss": 0.226,
      "step": 635
    },
    {
      "epoch": 0.6611226611226612,
      "grad_norm": 17.640363693237305,
      "learning_rate": 9.644259644259644e-06,
      "loss": 0.4552,
      "step": 636
    },
    {
      "epoch": 0.6621621621621622,
      "grad_norm": 17.001461029052734,
      "learning_rate": 9.641949641949642e-06,
      "loss": 0.5642,
      "step": 637
    },
    {
      "epoch": 0.6632016632016632,
      "grad_norm": 14.583857536315918,
      "learning_rate": 9.63963963963964e-06,
      "loss": 0.2794,
      "step": 638
    },
    {
      "epoch": 0.6642411642411642,
      "grad_norm": 16.399160385131836,
      "learning_rate": 9.637329637329637e-06,
      "loss": 0.5423,
      "step": 639
    },
    {
      "epoch": 0.6652806652806653,
      "grad_norm": 65.82854461669922,
      "learning_rate": 9.635019635019635e-06,
      "loss": 0.6206,
      "step": 640
    },
    {
      "epoch": 0.6663201663201663,
      "grad_norm": 0.8936071991920471,
      "learning_rate": 9.632709632709634e-06,
      "loss": 0.0147,
      "step": 641
    },
    {
      "epoch": 0.6673596673596673,
      "grad_norm": 8.22929859161377,
      "learning_rate": 9.630399630399632e-06,
      "loss": 0.1999,
      "step": 642
    },
    {
      "epoch": 0.6683991683991684,
      "grad_norm": 4.461599349975586,
      "learning_rate": 9.628089628089628e-06,
      "loss": 0.0999,
      "step": 643
    },
    {
      "epoch": 0.6694386694386695,
      "grad_norm": 6.589395046234131,
      "learning_rate": 9.625779625779626e-06,
      "loss": 0.1157,
      "step": 644
    },
    {
      "epoch": 0.6704781704781705,
      "grad_norm": 10.626900672912598,
      "learning_rate": 9.623469623469625e-06,
      "loss": 0.3657,
      "step": 645
    },
    {
      "epoch": 0.6715176715176715,
      "grad_norm": 21.5227108001709,
      "learning_rate": 9.621159621159621e-06,
      "loss": 0.9061,
      "step": 646
    },
    {
      "epoch": 0.6725571725571725,
      "grad_norm": 30.95146942138672,
      "learning_rate": 9.61884961884962e-06,
      "loss": 1.335,
      "step": 647
    },
    {
      "epoch": 0.6735966735966736,
      "grad_norm": 9.175050735473633,
      "learning_rate": 9.616539616539617e-06,
      "loss": 0.1422,
      "step": 648
    },
    {
      "epoch": 0.6746361746361746,
      "grad_norm": 14.263113975524902,
      "learning_rate": 9.614229614229616e-06,
      "loss": 1.1993,
      "step": 649
    },
    {
      "epoch": 0.6756756756756757,
      "grad_norm": 5.766702175140381,
      "learning_rate": 9.611919611919612e-06,
      "loss": 0.1147,
      "step": 650
    },
    {
      "epoch": 0.6767151767151767,
      "grad_norm": 10.668768882751465,
      "learning_rate": 9.60960960960961e-06,
      "loss": 0.4003,
      "step": 651
    },
    {
      "epoch": 0.6777546777546778,
      "grad_norm": 10.931303024291992,
      "learning_rate": 9.607299607299609e-06,
      "loss": 0.1814,
      "step": 652
    },
    {
      "epoch": 0.6787941787941788,
      "grad_norm": 4.092346668243408,
      "learning_rate": 9.604989604989607e-06,
      "loss": 0.1016,
      "step": 653
    },
    {
      "epoch": 0.6798336798336798,
      "grad_norm": 5.187844753265381,
      "learning_rate": 9.602679602679603e-06,
      "loss": 0.1261,
      "step": 654
    },
    {
      "epoch": 0.6808731808731808,
      "grad_norm": 21.333641052246094,
      "learning_rate": 9.600369600369601e-06,
      "loss": 0.782,
      "step": 655
    },
    {
      "epoch": 0.681912681912682,
      "grad_norm": 19.176406860351562,
      "learning_rate": 9.5980595980596e-06,
      "loss": 0.9153,
      "step": 656
    },
    {
      "epoch": 0.682952182952183,
      "grad_norm": 8.19583797454834,
      "learning_rate": 9.595749595749596e-06,
      "loss": 0.1717,
      "step": 657
    },
    {
      "epoch": 0.683991683991684,
      "grad_norm": 6.486777305603027,
      "learning_rate": 9.593439593439594e-06,
      "loss": 0.1231,
      "step": 658
    },
    {
      "epoch": 0.685031185031185,
      "grad_norm": 0.35394906997680664,
      "learning_rate": 9.591129591129592e-06,
      "loss": 0.0059,
      "step": 659
    },
    {
      "epoch": 0.6860706860706861,
      "grad_norm": 8.42319393157959,
      "learning_rate": 9.58881958881959e-06,
      "loss": 0.2822,
      "step": 660
    },
    {
      "epoch": 0.6871101871101871,
      "grad_norm": 19.60753059387207,
      "learning_rate": 9.586509586509587e-06,
      "loss": 0.6453,
      "step": 661
    },
    {
      "epoch": 0.6881496881496881,
      "grad_norm": 5.556179523468018,
      "learning_rate": 9.584199584199585e-06,
      "loss": 0.0812,
      "step": 662
    },
    {
      "epoch": 0.6891891891891891,
      "grad_norm": 11.209933280944824,
      "learning_rate": 9.581889581889584e-06,
      "loss": 0.3448,
      "step": 663
    },
    {
      "epoch": 0.6902286902286903,
      "grad_norm": 19.909692764282227,
      "learning_rate": 9.57957957957958e-06,
      "loss": 1.5944,
      "step": 664
    },
    {
      "epoch": 0.6912681912681913,
      "grad_norm": 11.300826072692871,
      "learning_rate": 9.577269577269578e-06,
      "loss": 0.4473,
      "step": 665
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 12.868157386779785,
      "learning_rate": 9.574959574959576e-06,
      "loss": 0.3579,
      "step": 666
    },
    {
      "epoch": 0.6933471933471933,
      "grad_norm": 13.064730644226074,
      "learning_rate": 9.572649572649575e-06,
      "loss": 0.5545,
      "step": 667
    },
    {
      "epoch": 0.6943866943866944,
      "grad_norm": 12.726256370544434,
      "learning_rate": 9.570339570339571e-06,
      "loss": 0.2416,
      "step": 668
    },
    {
      "epoch": 0.6954261954261954,
      "grad_norm": 5.459465980529785,
      "learning_rate": 9.56802956802957e-06,
      "loss": 0.0982,
      "step": 669
    },
    {
      "epoch": 0.6964656964656964,
      "grad_norm": 19.966798782348633,
      "learning_rate": 9.565719565719567e-06,
      "loss": 0.8197,
      "step": 670
    },
    {
      "epoch": 0.6975051975051975,
      "grad_norm": 12.257415771484375,
      "learning_rate": 9.563409563409564e-06,
      "loss": 0.6301,
      "step": 671
    },
    {
      "epoch": 0.6985446985446986,
      "grad_norm": 10.835104942321777,
      "learning_rate": 9.561099561099562e-06,
      "loss": 0.5715,
      "step": 672
    },
    {
      "epoch": 0.6995841995841996,
      "grad_norm": 14.587878227233887,
      "learning_rate": 9.55878955878956e-06,
      "loss": 0.2015,
      "step": 673
    },
    {
      "epoch": 0.7006237006237006,
      "grad_norm": 3.3721587657928467,
      "learning_rate": 9.556479556479557e-06,
      "loss": 0.0399,
      "step": 674
    },
    {
      "epoch": 0.7016632016632016,
      "grad_norm": 6.011256217956543,
      "learning_rate": 9.554169554169555e-06,
      "loss": 0.0979,
      "step": 675
    },
    {
      "epoch": 0.7027027027027027,
      "grad_norm": 28.068397521972656,
      "learning_rate": 9.551859551859553e-06,
      "loss": 0.4636,
      "step": 676
    },
    {
      "epoch": 0.7037422037422038,
      "grad_norm": 9.368738174438477,
      "learning_rate": 9.54954954954955e-06,
      "loss": 0.2166,
      "step": 677
    },
    {
      "epoch": 0.7047817047817048,
      "grad_norm": 13.36036205291748,
      "learning_rate": 9.547239547239548e-06,
      "loss": 0.3903,
      "step": 678
    },
    {
      "epoch": 0.7058212058212058,
      "grad_norm": 3.302922248840332,
      "learning_rate": 9.544929544929546e-06,
      "loss": 0.0396,
      "step": 679
    },
    {
      "epoch": 0.7068607068607069,
      "grad_norm": 8.110228538513184,
      "learning_rate": 9.542619542619543e-06,
      "loss": 0.1293,
      "step": 680
    },
    {
      "epoch": 0.7079002079002079,
      "grad_norm": 8.792784690856934,
      "learning_rate": 9.54030954030954e-06,
      "loss": 0.3419,
      "step": 681
    },
    {
      "epoch": 0.7089397089397089,
      "grad_norm": 8.643064498901367,
      "learning_rate": 9.537999537999539e-06,
      "loss": 0.2469,
      "step": 682
    },
    {
      "epoch": 0.7099792099792099,
      "grad_norm": 10.04169750213623,
      "learning_rate": 9.535689535689536e-06,
      "loss": 0.1695,
      "step": 683
    },
    {
      "epoch": 0.7110187110187111,
      "grad_norm": 6.619424343109131,
      "learning_rate": 9.533379533379534e-06,
      "loss": 0.2722,
      "step": 684
    },
    {
      "epoch": 0.7120582120582121,
      "grad_norm": 9.977747917175293,
      "learning_rate": 9.531069531069532e-06,
      "loss": 0.1084,
      "step": 685
    },
    {
      "epoch": 0.7130977130977131,
      "grad_norm": 7.454771518707275,
      "learning_rate": 9.528759528759528e-06,
      "loss": 0.1774,
      "step": 686
    },
    {
      "epoch": 0.7141372141372141,
      "grad_norm": 18.019685745239258,
      "learning_rate": 9.526449526449527e-06,
      "loss": 0.2778,
      "step": 687
    },
    {
      "epoch": 0.7151767151767152,
      "grad_norm": 15.221109390258789,
      "learning_rate": 9.524139524139525e-06,
      "loss": 0.6243,
      "step": 688
    },
    {
      "epoch": 0.7162162162162162,
      "grad_norm": 19.62790870666504,
      "learning_rate": 9.521829521829523e-06,
      "loss": 0.9515,
      "step": 689
    },
    {
      "epoch": 0.7172557172557172,
      "grad_norm": 13.727465629577637,
      "learning_rate": 9.51951951951952e-06,
      "loss": 0.2306,
      "step": 690
    },
    {
      "epoch": 0.7182952182952183,
      "grad_norm": 25.633102416992188,
      "learning_rate": 9.517209517209518e-06,
      "loss": 2.2608,
      "step": 691
    },
    {
      "epoch": 0.7193347193347194,
      "grad_norm": 7.654299736022949,
      "learning_rate": 9.514899514899516e-06,
      "loss": 0.0771,
      "step": 692
    },
    {
      "epoch": 0.7203742203742204,
      "grad_norm": 10.022356986999512,
      "learning_rate": 9.512589512589512e-06,
      "loss": 0.4251,
      "step": 693
    },
    {
      "epoch": 0.7214137214137214,
      "grad_norm": 7.501932621002197,
      "learning_rate": 9.51027951027951e-06,
      "loss": 0.2412,
      "step": 694
    },
    {
      "epoch": 0.7224532224532224,
      "grad_norm": 9.365520477294922,
      "learning_rate": 9.507969507969509e-06,
      "loss": 0.2679,
      "step": 695
    },
    {
      "epoch": 0.7234927234927235,
      "grad_norm": 1.676113247871399,
      "learning_rate": 9.505659505659507e-06,
      "loss": 0.0282,
      "step": 696
    },
    {
      "epoch": 0.7245322245322245,
      "grad_norm": 0.15936121344566345,
      "learning_rate": 9.503349503349503e-06,
      "loss": 0.0032,
      "step": 697
    },
    {
      "epoch": 0.7255717255717256,
      "grad_norm": 6.274758338928223,
      "learning_rate": 9.501039501039502e-06,
      "loss": 0.1157,
      "step": 698
    },
    {
      "epoch": 0.7266112266112266,
      "grad_norm": 16.684329986572266,
      "learning_rate": 9.4987294987295e-06,
      "loss": 1.1177,
      "step": 699
    },
    {
      "epoch": 0.7276507276507277,
      "grad_norm": 16.026220321655273,
      "learning_rate": 9.496419496419496e-06,
      "loss": 0.7498,
      "step": 700
    },
    {
      "epoch": 0.7286902286902287,
      "grad_norm": 9.84882640838623,
      "learning_rate": 9.494109494109494e-06,
      "loss": 0.2845,
      "step": 701
    },
    {
      "epoch": 0.7297297297297297,
      "grad_norm": 18.592073440551758,
      "learning_rate": 9.491799491799493e-06,
      "loss": 1.047,
      "step": 702
    },
    {
      "epoch": 0.7307692307692307,
      "grad_norm": 1.2101701498031616,
      "learning_rate": 9.489489489489491e-06,
      "loss": 0.0184,
      "step": 703
    },
    {
      "epoch": 0.7318087318087318,
      "grad_norm": 1.962371826171875,
      "learning_rate": 9.487179487179487e-06,
      "loss": 0.0255,
      "step": 704
    },
    {
      "epoch": 0.7328482328482329,
      "grad_norm": 8.123842239379883,
      "learning_rate": 9.484869484869486e-06,
      "loss": 0.1329,
      "step": 705
    },
    {
      "epoch": 0.7338877338877339,
      "grad_norm": 0.8697485327720642,
      "learning_rate": 9.482559482559484e-06,
      "loss": 0.008,
      "step": 706
    },
    {
      "epoch": 0.7349272349272349,
      "grad_norm": 12.17270278930664,
      "learning_rate": 9.480249480249482e-06,
      "loss": 0.3108,
      "step": 707
    },
    {
      "epoch": 0.735966735966736,
      "grad_norm": 10.54183578491211,
      "learning_rate": 9.477939477939478e-06,
      "loss": 0.4662,
      "step": 708
    },
    {
      "epoch": 0.737006237006237,
      "grad_norm": 28.778478622436523,
      "learning_rate": 9.475629475629477e-06,
      "loss": 0.3674,
      "step": 709
    },
    {
      "epoch": 0.738045738045738,
      "grad_norm": 7.773111820220947,
      "learning_rate": 9.473319473319475e-06,
      "loss": 0.2856,
      "step": 710
    },
    {
      "epoch": 0.739085239085239,
      "grad_norm": 4.889204502105713,
      "learning_rate": 9.471009471009471e-06,
      "loss": 0.0689,
      "step": 711
    },
    {
      "epoch": 0.7401247401247402,
      "grad_norm": 11.898750305175781,
      "learning_rate": 9.46869946869947e-06,
      "loss": 0.505,
      "step": 712
    },
    {
      "epoch": 0.7411642411642412,
      "grad_norm": 1.3363757133483887,
      "learning_rate": 9.466389466389468e-06,
      "loss": 0.021,
      "step": 713
    },
    {
      "epoch": 0.7422037422037422,
      "grad_norm": 9.92277717590332,
      "learning_rate": 9.464079464079466e-06,
      "loss": 0.2794,
      "step": 714
    },
    {
      "epoch": 0.7432432432432432,
      "grad_norm": 16.263818740844727,
      "learning_rate": 9.461769461769462e-06,
      "loss": 1.2749,
      "step": 715
    },
    {
      "epoch": 0.7442827442827443,
      "grad_norm": 4.112827777862549,
      "learning_rate": 9.45945945945946e-06,
      "loss": 0.0872,
      "step": 716
    },
    {
      "epoch": 0.7453222453222453,
      "grad_norm": 12.491456031799316,
      "learning_rate": 9.457149457149459e-06,
      "loss": 0.7282,
      "step": 717
    },
    {
      "epoch": 0.7463617463617463,
      "grad_norm": 8.743971824645996,
      "learning_rate": 9.454839454839455e-06,
      "loss": 0.1595,
      "step": 718
    },
    {
      "epoch": 0.7474012474012474,
      "grad_norm": 3.7295806407928467,
      "learning_rate": 9.452529452529453e-06,
      "loss": 0.0636,
      "step": 719
    },
    {
      "epoch": 0.7484407484407485,
      "grad_norm": 10.543453216552734,
      "learning_rate": 9.450219450219452e-06,
      "loss": 0.1802,
      "step": 720
    },
    {
      "epoch": 0.7494802494802495,
      "grad_norm": 3.609959840774536,
      "learning_rate": 9.44790944790945e-06,
      "loss": 0.1445,
      "step": 721
    },
    {
      "epoch": 0.7505197505197505,
      "grad_norm": 1.9326978921890259,
      "learning_rate": 9.445599445599446e-06,
      "loss": 0.0255,
      "step": 722
    },
    {
      "epoch": 0.7515592515592515,
      "grad_norm": 8.692984580993652,
      "learning_rate": 9.443289443289445e-06,
      "loss": 0.5295,
      "step": 723
    },
    {
      "epoch": 0.7525987525987526,
      "grad_norm": 13.966374397277832,
      "learning_rate": 9.440979440979443e-06,
      "loss": 0.5148,
      "step": 724
    },
    {
      "epoch": 0.7536382536382537,
      "grad_norm": 5.916383743286133,
      "learning_rate": 9.43866943866944e-06,
      "loss": 0.1337,
      "step": 725
    },
    {
      "epoch": 0.7546777546777547,
      "grad_norm": 0.8488734364509583,
      "learning_rate": 9.436359436359437e-06,
      "loss": 0.0174,
      "step": 726
    },
    {
      "epoch": 0.7557172557172557,
      "grad_norm": 0.8473299145698547,
      "learning_rate": 9.434049434049436e-06,
      "loss": 0.016,
      "step": 727
    },
    {
      "epoch": 0.7567567567567568,
      "grad_norm": 0.2719501554965973,
      "learning_rate": 9.431739431739432e-06,
      "loss": 0.0036,
      "step": 728
    },
    {
      "epoch": 0.7577962577962578,
      "grad_norm": 12.92343807220459,
      "learning_rate": 9.42942942942943e-06,
      "loss": 0.5638,
      "step": 729
    },
    {
      "epoch": 0.7588357588357588,
      "grad_norm": 5.92317533493042,
      "learning_rate": 9.427119427119428e-06,
      "loss": 0.1339,
      "step": 730
    },
    {
      "epoch": 0.7598752598752598,
      "grad_norm": 14.713382720947266,
      "learning_rate": 9.424809424809425e-06,
      "loss": 0.3023,
      "step": 731
    },
    {
      "epoch": 0.760914760914761,
      "grad_norm": 8.398733139038086,
      "learning_rate": 9.422499422499423e-06,
      "loss": 0.3296,
      "step": 732
    },
    {
      "epoch": 0.761954261954262,
      "grad_norm": 7.095760822296143,
      "learning_rate": 9.420189420189421e-06,
      "loss": 0.1501,
      "step": 733
    },
    {
      "epoch": 0.762993762993763,
      "grad_norm": 9.981751441955566,
      "learning_rate": 9.417879417879418e-06,
      "loss": 0.5741,
      "step": 734
    },
    {
      "epoch": 0.764033264033264,
      "grad_norm": 0.5671398043632507,
      "learning_rate": 9.415569415569416e-06,
      "loss": 0.0087,
      "step": 735
    },
    {
      "epoch": 0.7650727650727651,
      "grad_norm": 17.42526626586914,
      "learning_rate": 9.413259413259414e-06,
      "loss": 0.6538,
      "step": 736
    },
    {
      "epoch": 0.7661122661122661,
      "grad_norm": 8.388923645019531,
      "learning_rate": 9.41094941094941e-06,
      "loss": 0.2159,
      "step": 737
    },
    {
      "epoch": 0.7671517671517671,
      "grad_norm": 14.789589881896973,
      "learning_rate": 9.408639408639409e-06,
      "loss": 0.2748,
      "step": 738
    },
    {
      "epoch": 0.7681912681912682,
      "grad_norm": 12.18989086151123,
      "learning_rate": 9.406329406329407e-06,
      "loss": 0.6697,
      "step": 739
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 21.005599975585938,
      "learning_rate": 9.404019404019404e-06,
      "loss": 1.4857,
      "step": 740
    },
    {
      "epoch": 0.7702702702702703,
      "grad_norm": 4.1699748039245605,
      "learning_rate": 9.401709401709402e-06,
      "loss": 0.0761,
      "step": 741
    },
    {
      "epoch": 0.7713097713097713,
      "grad_norm": 16.020122528076172,
      "learning_rate": 9.3993993993994e-06,
      "loss": 0.5824,
      "step": 742
    },
    {
      "epoch": 0.7723492723492723,
      "grad_norm": 14.067607879638672,
      "learning_rate": 9.397089397089398e-06,
      "loss": 0.4511,
      "step": 743
    },
    {
      "epoch": 0.7733887733887734,
      "grad_norm": 8.97590160369873,
      "learning_rate": 9.394779394779395e-06,
      "loss": 0.2776,
      "step": 744
    },
    {
      "epoch": 0.7744282744282744,
      "grad_norm": 5.633431911468506,
      "learning_rate": 9.392469392469393e-06,
      "loss": 0.1346,
      "step": 745
    },
    {
      "epoch": 0.7754677754677755,
      "grad_norm": 12.008380889892578,
      "learning_rate": 9.390159390159391e-06,
      "loss": 0.3179,
      "step": 746
    },
    {
      "epoch": 0.7765072765072765,
      "grad_norm": 2.2428348064422607,
      "learning_rate": 9.387849387849388e-06,
      "loss": 0.0619,
      "step": 747
    },
    {
      "epoch": 0.7775467775467776,
      "grad_norm": 17.27653694152832,
      "learning_rate": 9.385539385539386e-06,
      "loss": 0.814,
      "step": 748
    },
    {
      "epoch": 0.7785862785862786,
      "grad_norm": 4.97644567489624,
      "learning_rate": 9.383229383229384e-06,
      "loss": 0.0552,
      "step": 749
    },
    {
      "epoch": 0.7796257796257796,
      "grad_norm": 3.5207571983337402,
      "learning_rate": 9.380919380919382e-06,
      "loss": 0.038,
      "step": 750
    },
    {
      "epoch": 0.7806652806652806,
      "grad_norm": 4.021271228790283,
      "learning_rate": 9.378609378609379e-06,
      "loss": 0.0547,
      "step": 751
    },
    {
      "epoch": 0.7817047817047817,
      "grad_norm": 11.576969146728516,
      "learning_rate": 9.376299376299377e-06,
      "loss": 0.948,
      "step": 752
    },
    {
      "epoch": 0.7827442827442828,
      "grad_norm": 0.44165757298469543,
      "learning_rate": 9.373989373989375e-06,
      "loss": 0.0109,
      "step": 753
    },
    {
      "epoch": 0.7837837837837838,
      "grad_norm": 7.726773262023926,
      "learning_rate": 9.371679371679371e-06,
      "loss": 0.2671,
      "step": 754
    },
    {
      "epoch": 0.7848232848232848,
      "grad_norm": 8.891253471374512,
      "learning_rate": 9.36936936936937e-06,
      "loss": 0.2311,
      "step": 755
    },
    {
      "epoch": 0.7858627858627859,
      "grad_norm": 11.3536958694458,
      "learning_rate": 9.367059367059368e-06,
      "loss": 0.6166,
      "step": 756
    },
    {
      "epoch": 0.7869022869022869,
      "grad_norm": 11.843234062194824,
      "learning_rate": 9.364749364749366e-06,
      "loss": 0.7811,
      "step": 757
    },
    {
      "epoch": 0.7879417879417879,
      "grad_norm": 14.331742286682129,
      "learning_rate": 9.362439362439363e-06,
      "loss": 0.5065,
      "step": 758
    },
    {
      "epoch": 0.7889812889812889,
      "grad_norm": 23.064044952392578,
      "learning_rate": 9.36012936012936e-06,
      "loss": 0.9797,
      "step": 759
    },
    {
      "epoch": 0.7900207900207901,
      "grad_norm": 3.3058671951293945,
      "learning_rate": 9.357819357819359e-06,
      "loss": 0.0457,
      "step": 760
    },
    {
      "epoch": 0.7910602910602911,
      "grad_norm": 8.004167556762695,
      "learning_rate": 9.355509355509357e-06,
      "loss": 0.1693,
      "step": 761
    },
    {
      "epoch": 0.7920997920997921,
      "grad_norm": 10.097188949584961,
      "learning_rate": 9.353199353199354e-06,
      "loss": 0.247,
      "step": 762
    },
    {
      "epoch": 0.7931392931392931,
      "grad_norm": 5.271298408508301,
      "learning_rate": 9.350889350889352e-06,
      "loss": 0.0974,
      "step": 763
    },
    {
      "epoch": 0.7941787941787942,
      "grad_norm": 0.3096296787261963,
      "learning_rate": 9.34857934857935e-06,
      "loss": 0.0059,
      "step": 764
    },
    {
      "epoch": 0.7952182952182952,
      "grad_norm": 11.955662727355957,
      "learning_rate": 9.346269346269346e-06,
      "loss": 0.6279,
      "step": 765
    },
    {
      "epoch": 0.7962577962577962,
      "grad_norm": 4.01197624206543,
      "learning_rate": 9.343959343959345e-06,
      "loss": 0.0092,
      "step": 766
    },
    {
      "epoch": 0.7972972972972973,
      "grad_norm": 1.3834837675094604,
      "learning_rate": 9.341649341649343e-06,
      "loss": 0.0126,
      "step": 767
    },
    {
      "epoch": 0.7983367983367984,
      "grad_norm": 10.295829772949219,
      "learning_rate": 9.339339339339341e-06,
      "loss": 0.2918,
      "step": 768
    },
    {
      "epoch": 0.7993762993762994,
      "grad_norm": 5.462759494781494,
      "learning_rate": 9.337029337029338e-06,
      "loss": 0.1415,
      "step": 769
    },
    {
      "epoch": 0.8004158004158004,
      "grad_norm": 10.10068416595459,
      "learning_rate": 9.334719334719336e-06,
      "loss": 0.2449,
      "step": 770
    },
    {
      "epoch": 0.8014553014553014,
      "grad_norm": 13.373099327087402,
      "learning_rate": 9.332409332409334e-06,
      "loss": 0.2247,
      "step": 771
    },
    {
      "epoch": 0.8024948024948025,
      "grad_norm": 9.904016494750977,
      "learning_rate": 9.33009933009933e-06,
      "loss": 0.4769,
      "step": 772
    },
    {
      "epoch": 0.8035343035343036,
      "grad_norm": 5.0328898429870605,
      "learning_rate": 9.327789327789329e-06,
      "loss": 0.0871,
      "step": 773
    },
    {
      "epoch": 0.8045738045738046,
      "grad_norm": 2.7860841751098633,
      "learning_rate": 9.325479325479327e-06,
      "loss": 0.0474,
      "step": 774
    },
    {
      "epoch": 0.8056133056133056,
      "grad_norm": 1.0327810049057007,
      "learning_rate": 9.323169323169325e-06,
      "loss": 0.0124,
      "step": 775
    },
    {
      "epoch": 0.8066528066528067,
      "grad_norm": 20.533740997314453,
      "learning_rate": 9.320859320859322e-06,
      "loss": 0.5364,
      "step": 776
    },
    {
      "epoch": 0.8076923076923077,
      "grad_norm": 1.5147625207901,
      "learning_rate": 9.31854931854932e-06,
      "loss": 0.0465,
      "step": 777
    },
    {
      "epoch": 0.8087318087318087,
      "grad_norm": 2.101473331451416,
      "learning_rate": 9.316239316239318e-06,
      "loss": 0.042,
      "step": 778
    },
    {
      "epoch": 0.8097713097713097,
      "grad_norm": 13.39181137084961,
      "learning_rate": 9.313929313929314e-06,
      "loss": 0.3735,
      "step": 779
    },
    {
      "epoch": 0.8108108108108109,
      "grad_norm": 18.286285400390625,
      "learning_rate": 9.311619311619313e-06,
      "loss": 0.6235,
      "step": 780
    },
    {
      "epoch": 0.8118503118503119,
      "grad_norm": 19.759904861450195,
      "learning_rate": 9.30930930930931e-06,
      "loss": 0.722,
      "step": 781
    },
    {
      "epoch": 0.8128898128898129,
      "grad_norm": 12.798063278198242,
      "learning_rate": 9.306999306999307e-06,
      "loss": 0.3945,
      "step": 782
    },
    {
      "epoch": 0.8139293139293139,
      "grad_norm": 6.071016788482666,
      "learning_rate": 9.304689304689305e-06,
      "loss": 0.2655,
      "step": 783
    },
    {
      "epoch": 0.814968814968815,
      "grad_norm": 2.006742477416992,
      "learning_rate": 9.302379302379304e-06,
      "loss": 0.0316,
      "step": 784
    },
    {
      "epoch": 0.816008316008316,
      "grad_norm": 19.665082931518555,
      "learning_rate": 9.3000693000693e-06,
      "loss": 1.3005,
      "step": 785
    },
    {
      "epoch": 0.817047817047817,
      "grad_norm": 19.220882415771484,
      "learning_rate": 9.297759297759298e-06,
      "loss": 0.6756,
      "step": 786
    },
    {
      "epoch": 0.818087318087318,
      "grad_norm": 7.785876274108887,
      "learning_rate": 9.295449295449297e-06,
      "loss": 0.2178,
      "step": 787
    },
    {
      "epoch": 0.8191268191268192,
      "grad_norm": 11.736283302307129,
      "learning_rate": 9.293139293139293e-06,
      "loss": 0.1697,
      "step": 788
    },
    {
      "epoch": 0.8201663201663202,
      "grad_norm": 12.8834228515625,
      "learning_rate": 9.290829290829291e-06,
      "loss": 0.8898,
      "step": 789
    },
    {
      "epoch": 0.8212058212058212,
      "grad_norm": 7.315210819244385,
      "learning_rate": 9.28851928851929e-06,
      "loss": 0.1394,
      "step": 790
    },
    {
      "epoch": 0.8222453222453222,
      "grad_norm": 13.195050239562988,
      "learning_rate": 9.286209286209288e-06,
      "loss": 0.5235,
      "step": 791
    },
    {
      "epoch": 0.8232848232848233,
      "grad_norm": 4.913758754730225,
      "learning_rate": 9.283899283899284e-06,
      "loss": 0.1314,
      "step": 792
    },
    {
      "epoch": 0.8243243243243243,
      "grad_norm": 7.278382301330566,
      "learning_rate": 9.281589281589282e-06,
      "loss": 0.2362,
      "step": 793
    },
    {
      "epoch": 0.8253638253638254,
      "grad_norm": 8.85083293914795,
      "learning_rate": 9.27927927927928e-06,
      "loss": 0.236,
      "step": 794
    },
    {
      "epoch": 0.8264033264033264,
      "grad_norm": 11.661330223083496,
      "learning_rate": 9.276969276969277e-06,
      "loss": 0.4306,
      "step": 795
    },
    {
      "epoch": 0.8274428274428275,
      "grad_norm": 7.37078332901001,
      "learning_rate": 9.274659274659275e-06,
      "loss": 0.1825,
      "step": 796
    },
    {
      "epoch": 0.8284823284823285,
      "grad_norm": 12.864847183227539,
      "learning_rate": 9.272349272349273e-06,
      "loss": 0.6644,
      "step": 797
    },
    {
      "epoch": 0.8295218295218295,
      "grad_norm": 16.365703582763672,
      "learning_rate": 9.27003927003927e-06,
      "loss": 0.9907,
      "step": 798
    },
    {
      "epoch": 0.8305613305613305,
      "grad_norm": 14.491663932800293,
      "learning_rate": 9.267729267729268e-06,
      "loss": 0.574,
      "step": 799
    },
    {
      "epoch": 0.8316008316008316,
      "grad_norm": 6.973012447357178,
      "learning_rate": 9.265419265419266e-06,
      "loss": 0.2341,
      "step": 800
    },
    {
      "epoch": 0.8326403326403327,
      "grad_norm": 0.363965779542923,
      "learning_rate": 9.263109263109263e-06,
      "loss": 0.0068,
      "step": 801
    },
    {
      "epoch": 0.8336798336798337,
      "grad_norm": 0.5364904999732971,
      "learning_rate": 9.260799260799261e-06,
      "loss": 0.0132,
      "step": 802
    },
    {
      "epoch": 0.8347193347193347,
      "grad_norm": 2.5990946292877197,
      "learning_rate": 9.258489258489259e-06,
      "loss": 0.0541,
      "step": 803
    },
    {
      "epoch": 0.8357588357588358,
      "grad_norm": 4.446463108062744,
      "learning_rate": 9.256179256179257e-06,
      "loss": 0.093,
      "step": 804
    },
    {
      "epoch": 0.8367983367983368,
      "grad_norm": 11.481074333190918,
      "learning_rate": 9.253869253869254e-06,
      "loss": 0.8781,
      "step": 805
    },
    {
      "epoch": 0.8378378378378378,
      "grad_norm": 5.719520092010498,
      "learning_rate": 9.251559251559252e-06,
      "loss": 0.1585,
      "step": 806
    },
    {
      "epoch": 0.8388773388773388,
      "grad_norm": 0.9901307821273804,
      "learning_rate": 9.24924924924925e-06,
      "loss": 0.0226,
      "step": 807
    },
    {
      "epoch": 0.83991683991684,
      "grad_norm": 6.784252166748047,
      "learning_rate": 9.246939246939247e-06,
      "loss": 0.1927,
      "step": 808
    },
    {
      "epoch": 0.840956340956341,
      "grad_norm": 0.0940762609243393,
      "learning_rate": 9.244629244629245e-06,
      "loss": 0.0017,
      "step": 809
    },
    {
      "epoch": 0.841995841995842,
      "grad_norm": 14.0625581741333,
      "learning_rate": 9.242319242319243e-06,
      "loss": 0.6236,
      "step": 810
    },
    {
      "epoch": 0.843035343035343,
      "grad_norm": 3.779634714126587,
      "learning_rate": 9.240009240009241e-06,
      "loss": 0.1269,
      "step": 811
    },
    {
      "epoch": 0.8440748440748441,
      "grad_norm": 8.7300443649292,
      "learning_rate": 9.237699237699238e-06,
      "loss": 0.1096,
      "step": 812
    },
    {
      "epoch": 0.8451143451143451,
      "grad_norm": 4.038877010345459,
      "learning_rate": 9.235389235389236e-06,
      "loss": 0.0703,
      "step": 813
    },
    {
      "epoch": 0.8461538461538461,
      "grad_norm": 22.447866439819336,
      "learning_rate": 9.233079233079234e-06,
      "loss": 1.1194,
      "step": 814
    },
    {
      "epoch": 0.8471933471933472,
      "grad_norm": 5.937042713165283,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.2543,
      "step": 815
    },
    {
      "epoch": 0.8482328482328483,
      "grad_norm": 0.08702496439218521,
      "learning_rate": 9.228459228459229e-06,
      "loss": 0.0019,
      "step": 816
    },
    {
      "epoch": 0.8492723492723493,
      "grad_norm": 9.366803169250488,
      "learning_rate": 9.226149226149227e-06,
      "loss": 0.3223,
      "step": 817
    },
    {
      "epoch": 0.8503118503118503,
      "grad_norm": 6.609611988067627,
      "learning_rate": 9.223839223839225e-06,
      "loss": 0.1015,
      "step": 818
    },
    {
      "epoch": 0.8513513513513513,
      "grad_norm": 5.29298210144043,
      "learning_rate": 9.221529221529222e-06,
      "loss": 0.3793,
      "step": 819
    },
    {
      "epoch": 0.8523908523908524,
      "grad_norm": 3.444406509399414,
      "learning_rate": 9.21921921921922e-06,
      "loss": 0.1107,
      "step": 820
    },
    {
      "epoch": 0.8534303534303534,
      "grad_norm": 1.0146065950393677,
      "learning_rate": 9.216909216909218e-06,
      "loss": 0.0188,
      "step": 821
    },
    {
      "epoch": 0.8544698544698545,
      "grad_norm": 0.4745628833770752,
      "learning_rate": 9.214599214599216e-06,
      "loss": 0.0088,
      "step": 822
    },
    {
      "epoch": 0.8555093555093555,
      "grad_norm": 16.4841365814209,
      "learning_rate": 9.212289212289213e-06,
      "loss": 0.7025,
      "step": 823
    },
    {
      "epoch": 0.8565488565488566,
      "grad_norm": 5.551940441131592,
      "learning_rate": 9.209979209979211e-06,
      "loss": 0.0577,
      "step": 824
    },
    {
      "epoch": 0.8575883575883576,
      "grad_norm": 5.3550262451171875,
      "learning_rate": 9.207669207669209e-06,
      "loss": 0.0957,
      "step": 825
    },
    {
      "epoch": 0.8586278586278586,
      "grad_norm": 18.20002555847168,
      "learning_rate": 9.205359205359206e-06,
      "loss": 1.2307,
      "step": 826
    },
    {
      "epoch": 0.8596673596673596,
      "grad_norm": 3.5077757835388184,
      "learning_rate": 9.203049203049204e-06,
      "loss": 0.0976,
      "step": 827
    },
    {
      "epoch": 0.8607068607068608,
      "grad_norm": 0.0986674576997757,
      "learning_rate": 9.200739200739202e-06,
      "loss": 0.003,
      "step": 828
    },
    {
      "epoch": 0.8617463617463618,
      "grad_norm": 14.541893005371094,
      "learning_rate": 9.1984291984292e-06,
      "loss": 0.5262,
      "step": 829
    },
    {
      "epoch": 0.8627858627858628,
      "grad_norm": 10.445574760437012,
      "learning_rate": 9.196119196119197e-06,
      "loss": 0.3528,
      "step": 830
    },
    {
      "epoch": 0.8638253638253638,
      "grad_norm": 9.452934265136719,
      "learning_rate": 9.193809193809195e-06,
      "loss": 0.2406,
      "step": 831
    },
    {
      "epoch": 0.8648648648648649,
      "grad_norm": 10.794607162475586,
      "learning_rate": 9.191499191499193e-06,
      "loss": 0.2028,
      "step": 832
    },
    {
      "epoch": 0.8659043659043659,
      "grad_norm": 11.823083877563477,
      "learning_rate": 9.189189189189191e-06,
      "loss": 0.5005,
      "step": 833
    },
    {
      "epoch": 0.8669438669438669,
      "grad_norm": 12.307615280151367,
      "learning_rate": 9.186879186879188e-06,
      "loss": 0.2977,
      "step": 834
    },
    {
      "epoch": 0.867983367983368,
      "grad_norm": 12.70445728302002,
      "learning_rate": 9.184569184569186e-06,
      "loss": 0.5725,
      "step": 835
    },
    {
      "epoch": 0.8690228690228691,
      "grad_norm": 2.4809796810150146,
      "learning_rate": 9.182259182259184e-06,
      "loss": 0.0758,
      "step": 836
    },
    {
      "epoch": 0.8700623700623701,
      "grad_norm": 1.8290390968322754,
      "learning_rate": 9.17994917994918e-06,
      "loss": 0.0356,
      "step": 837
    },
    {
      "epoch": 0.8711018711018711,
      "grad_norm": 16.160789489746094,
      "learning_rate": 9.177639177639179e-06,
      "loss": 1.0367,
      "step": 838
    },
    {
      "epoch": 0.8721413721413721,
      "grad_norm": 14.823076248168945,
      "learning_rate": 9.175329175329177e-06,
      "loss": 0.603,
      "step": 839
    },
    {
      "epoch": 0.8731808731808732,
      "grad_norm": 16.44916534423828,
      "learning_rate": 9.173019173019174e-06,
      "loss": 0.2872,
      "step": 840
    },
    {
      "epoch": 0.8742203742203742,
      "grad_norm": 9.719271659851074,
      "learning_rate": 9.170709170709172e-06,
      "loss": 0.3966,
      "step": 841
    },
    {
      "epoch": 0.8752598752598753,
      "grad_norm": 0.07257251441478729,
      "learning_rate": 9.16839916839917e-06,
      "loss": 0.0012,
      "step": 842
    },
    {
      "epoch": 0.8762993762993763,
      "grad_norm": 7.706364631652832,
      "learning_rate": 9.166089166089166e-06,
      "loss": 0.2788,
      "step": 843
    },
    {
      "epoch": 0.8773388773388774,
      "grad_norm": 7.656033992767334,
      "learning_rate": 9.163779163779165e-06,
      "loss": 0.1487,
      "step": 844
    },
    {
      "epoch": 0.8783783783783784,
      "grad_norm": 2.814105749130249,
      "learning_rate": 9.161469161469163e-06,
      "loss": 0.031,
      "step": 845
    },
    {
      "epoch": 0.8794178794178794,
      "grad_norm": 7.749004364013672,
      "learning_rate": 9.15915915915916e-06,
      "loss": 0.0998,
      "step": 846
    },
    {
      "epoch": 0.8804573804573804,
      "grad_norm": 1.6006381511688232,
      "learning_rate": 9.156849156849157e-06,
      "loss": 0.0176,
      "step": 847
    },
    {
      "epoch": 0.8814968814968815,
      "grad_norm": 4.9961419105529785,
      "learning_rate": 9.154539154539156e-06,
      "loss": 0.0861,
      "step": 848
    },
    {
      "epoch": 0.8825363825363826,
      "grad_norm": 12.169288635253906,
      "learning_rate": 9.152229152229152e-06,
      "loss": 0.2653,
      "step": 849
    },
    {
      "epoch": 0.8835758835758836,
      "grad_norm": 14.644551277160645,
      "learning_rate": 9.14991914991915e-06,
      "loss": 0.5879,
      "step": 850
    },
    {
      "epoch": 0.8846153846153846,
      "grad_norm": 13.133461952209473,
      "learning_rate": 9.147609147609149e-06,
      "loss": 0.3398,
      "step": 851
    },
    {
      "epoch": 0.8856548856548857,
      "grad_norm": 8.79302978515625,
      "learning_rate": 9.145299145299145e-06,
      "loss": 0.3166,
      "step": 852
    },
    {
      "epoch": 0.8866943866943867,
      "grad_norm": 0.8859695792198181,
      "learning_rate": 9.142989142989143e-06,
      "loss": 0.0172,
      "step": 853
    },
    {
      "epoch": 0.8877338877338877,
      "grad_norm": 11.242350578308105,
      "learning_rate": 9.140679140679141e-06,
      "loss": 0.4063,
      "step": 854
    },
    {
      "epoch": 0.8887733887733887,
      "grad_norm": 9.41787052154541,
      "learning_rate": 9.138369138369138e-06,
      "loss": 0.7275,
      "step": 855
    },
    {
      "epoch": 0.8898128898128899,
      "grad_norm": 6.311247825622559,
      "learning_rate": 9.136059136059136e-06,
      "loss": 0.2133,
      "step": 856
    },
    {
      "epoch": 0.8908523908523909,
      "grad_norm": 9.024970054626465,
      "learning_rate": 9.133749133749134e-06,
      "loss": 0.2356,
      "step": 857
    },
    {
      "epoch": 0.8918918918918919,
      "grad_norm": 12.511112213134766,
      "learning_rate": 9.131439131439132e-06,
      "loss": 0.6547,
      "step": 858
    },
    {
      "epoch": 0.8929313929313929,
      "grad_norm": 19.817020416259766,
      "learning_rate": 9.129129129129129e-06,
      "loss": 0.9193,
      "step": 859
    },
    {
      "epoch": 0.893970893970894,
      "grad_norm": 8.149579048156738,
      "learning_rate": 9.126819126819127e-06,
      "loss": 0.0618,
      "step": 860
    },
    {
      "epoch": 0.895010395010395,
      "grad_norm": 17.640111923217773,
      "learning_rate": 9.124509124509125e-06,
      "loss": 1.6197,
      "step": 861
    },
    {
      "epoch": 0.896049896049896,
      "grad_norm": 13.725643157958984,
      "learning_rate": 9.122199122199122e-06,
      "loss": 0.2125,
      "step": 862
    },
    {
      "epoch": 0.8970893970893971,
      "grad_norm": 13.83113956451416,
      "learning_rate": 9.11988911988912e-06,
      "loss": 0.3896,
      "step": 863
    },
    {
      "epoch": 0.8981288981288982,
      "grad_norm": 9.916857719421387,
      "learning_rate": 9.117579117579118e-06,
      "loss": 0.3564,
      "step": 864
    },
    {
      "epoch": 0.8991683991683992,
      "grad_norm": 20.230899810791016,
      "learning_rate": 9.115269115269116e-06,
      "loss": 0.4534,
      "step": 865
    },
    {
      "epoch": 0.9002079002079002,
      "grad_norm": 8.180204391479492,
      "learning_rate": 9.112959112959113e-06,
      "loss": 0.2119,
      "step": 866
    },
    {
      "epoch": 0.9012474012474012,
      "grad_norm": 14.04716968536377,
      "learning_rate": 9.110649110649111e-06,
      "loss": 0.936,
      "step": 867
    },
    {
      "epoch": 0.9022869022869023,
      "grad_norm": NaN,
      "learning_rate": 9.10833910833911e-06,
      "loss": 0.0104,
      "step": 868
    },
    {
      "epoch": 0.9033264033264033,
      "grad_norm": 3.089855909347534,
      "learning_rate": 9.106029106029107e-06,
      "loss": 0.0438,
      "step": 869
    },
    {
      "epoch": 0.9043659043659044,
      "grad_norm": 10.145646095275879,
      "learning_rate": 9.103719103719104e-06,
      "loss": 0.3031,
      "step": 870
    },
    {
      "epoch": 0.9054054054054054,
      "grad_norm": 7.578200817108154,
      "learning_rate": 9.101409101409102e-06,
      "loss": 0.2943,
      "step": 871
    },
    {
      "epoch": 0.9064449064449065,
      "grad_norm": 3.3656668663024902,
      "learning_rate": 9.0990990990991e-06,
      "loss": 0.0814,
      "step": 872
    },
    {
      "epoch": 0.9074844074844075,
      "grad_norm": 4.186402320861816,
      "learning_rate": 9.096789096789097e-06,
      "loss": 0.0662,
      "step": 873
    },
    {
      "epoch": 0.9085239085239085,
      "grad_norm": 13.625324249267578,
      "learning_rate": 9.094479094479095e-06,
      "loss": 2.2787,
      "step": 874
    },
    {
      "epoch": 0.9095634095634095,
      "grad_norm": 12.998013496398926,
      "learning_rate": 9.092169092169093e-06,
      "loss": 0.5581,
      "step": 875
    },
    {
      "epoch": 0.9106029106029107,
      "grad_norm": 17.50330352783203,
      "learning_rate": 9.089859089859091e-06,
      "loss": 0.9294,
      "step": 876
    },
    {
      "epoch": 0.9116424116424117,
      "grad_norm": 7.164412975311279,
      "learning_rate": 9.087549087549088e-06,
      "loss": 0.1759,
      "step": 877
    },
    {
      "epoch": 0.9126819126819127,
      "grad_norm": 5.057459354400635,
      "learning_rate": 9.085239085239086e-06,
      "loss": 0.0994,
      "step": 878
    },
    {
      "epoch": 0.9137214137214137,
      "grad_norm": 12.146804809570312,
      "learning_rate": 9.082929082929084e-06,
      "loss": 0.5214,
      "step": 879
    },
    {
      "epoch": 0.9147609147609148,
      "grad_norm": 1.3866074085235596,
      "learning_rate": 9.08061908061908e-06,
      "loss": 0.0279,
      "step": 880
    },
    {
      "epoch": 0.9158004158004158,
      "grad_norm": 6.697347164154053,
      "learning_rate": 9.078309078309079e-06,
      "loss": 0.142,
      "step": 881
    },
    {
      "epoch": 0.9168399168399168,
      "grad_norm": 10.428801536560059,
      "learning_rate": 9.075999075999077e-06,
      "loss": 0.4985,
      "step": 882
    },
    {
      "epoch": 0.9178794178794178,
      "grad_norm": 10.876057624816895,
      "learning_rate": 9.073689073689075e-06,
      "loss": 0.4017,
      "step": 883
    },
    {
      "epoch": 0.918918918918919,
      "grad_norm": 7.627023696899414,
      "learning_rate": 9.071379071379072e-06,
      "loss": 0.3287,
      "step": 884
    },
    {
      "epoch": 0.91995841995842,
      "grad_norm": 12.741010665893555,
      "learning_rate": 9.06906906906907e-06,
      "loss": 0.4065,
      "step": 885
    },
    {
      "epoch": 0.920997920997921,
      "grad_norm": 12.223453521728516,
      "learning_rate": 9.066759066759068e-06,
      "loss": 0.5238,
      "step": 886
    },
    {
      "epoch": 0.922037422037422,
      "grad_norm": 11.803918838500977,
      "learning_rate": 9.064449064449066e-06,
      "loss": 0.5711,
      "step": 887
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 5.423978805541992,
      "learning_rate": 9.062139062139063e-06,
      "loss": 0.1786,
      "step": 888
    },
    {
      "epoch": 0.9241164241164241,
      "grad_norm": 12.600936889648438,
      "learning_rate": 9.059829059829061e-06,
      "loss": 0.4155,
      "step": 889
    },
    {
      "epoch": 0.9251559251559252,
      "grad_norm": 5.410848140716553,
      "learning_rate": 9.05751905751906e-06,
      "loss": 0.121,
      "step": 890
    },
    {
      "epoch": 0.9261954261954262,
      "grad_norm": 1.709598422050476,
      "learning_rate": 9.055209055209056e-06,
      "loss": 0.0347,
      "step": 891
    },
    {
      "epoch": 0.9272349272349273,
      "grad_norm": 10.025089263916016,
      "learning_rate": 9.052899052899054e-06,
      "loss": 0.5089,
      "step": 892
    },
    {
      "epoch": 0.9282744282744283,
      "grad_norm": 0.5808941721916199,
      "learning_rate": 9.050589050589052e-06,
      "loss": 0.0147,
      "step": 893
    },
    {
      "epoch": 0.9293139293139293,
      "grad_norm": 11.988170623779297,
      "learning_rate": 9.048279048279049e-06,
      "loss": 0.9453,
      "step": 894
    },
    {
      "epoch": 0.9303534303534303,
      "grad_norm": 13.248468399047852,
      "learning_rate": 9.045969045969047e-06,
      "loss": 0.906,
      "step": 895
    },
    {
      "epoch": 0.9313929313929314,
      "grad_norm": 9.587570190429688,
      "learning_rate": 9.043659043659045e-06,
      "loss": 0.2799,
      "step": 896
    },
    {
      "epoch": 0.9324324324324325,
      "grad_norm": 4.256852626800537,
      "learning_rate": 9.041349041349042e-06,
      "loss": 0.0803,
      "step": 897
    },
    {
      "epoch": 0.9334719334719335,
      "grad_norm": 1.7316964864730835,
      "learning_rate": 9.03903903903904e-06,
      "loss": 0.0563,
      "step": 898
    },
    {
      "epoch": 0.9345114345114345,
      "grad_norm": 0.29858797788619995,
      "learning_rate": 9.036729036729038e-06,
      "loss": 0.0072,
      "step": 899
    },
    {
      "epoch": 0.9355509355509356,
      "grad_norm": 33.127559661865234,
      "learning_rate": 9.034419034419034e-06,
      "loss": 0.3419,
      "step": 900
    },
    {
      "epoch": 0.9365904365904366,
      "grad_norm": 3.9960570335388184,
      "learning_rate": 9.032109032109033e-06,
      "loss": 0.0265,
      "step": 901
    },
    {
      "epoch": 0.9376299376299376,
      "grad_norm": 5.864764213562012,
      "learning_rate": 9.029799029799031e-06,
      "loss": 0.1599,
      "step": 902
    },
    {
      "epoch": 0.9386694386694386,
      "grad_norm": 4.86962890625,
      "learning_rate": 9.027489027489027e-06,
      "loss": 0.2822,
      "step": 903
    },
    {
      "epoch": 0.9397089397089398,
      "grad_norm": 0.07891742140054703,
      "learning_rate": 9.025179025179026e-06,
      "loss": 0.002,
      "step": 904
    },
    {
      "epoch": 0.9407484407484408,
      "grad_norm": 1.544185996055603,
      "learning_rate": 9.022869022869024e-06,
      "loss": 0.025,
      "step": 905
    },
    {
      "epoch": 0.9417879417879418,
      "grad_norm": 18.113237380981445,
      "learning_rate": 9.02055902055902e-06,
      "loss": 1.8649,
      "step": 906
    },
    {
      "epoch": 0.9428274428274428,
      "grad_norm": 5.277674198150635,
      "learning_rate": 9.018249018249018e-06,
      "loss": 0.1455,
      "step": 907
    },
    {
      "epoch": 0.9438669438669439,
      "grad_norm": 3.8666481971740723,
      "learning_rate": 9.015939015939017e-06,
      "loss": 0.0412,
      "step": 908
    },
    {
      "epoch": 0.9449064449064449,
      "grad_norm": 5.831012725830078,
      "learning_rate": 9.013629013629013e-06,
      "loss": 0.5328,
      "step": 909
    },
    {
      "epoch": 0.9459459459459459,
      "grad_norm": 10.758270263671875,
      "learning_rate": 9.011319011319011e-06,
      "loss": 0.655,
      "step": 910
    },
    {
      "epoch": 0.946985446985447,
      "grad_norm": 6.609604835510254,
      "learning_rate": 9.00900900900901e-06,
      "loss": 0.2318,
      "step": 911
    },
    {
      "epoch": 0.9480249480249481,
      "grad_norm": 1.2616462707519531,
      "learning_rate": 9.006699006699008e-06,
      "loss": 0.0162,
      "step": 912
    },
    {
      "epoch": 0.9490644490644491,
      "grad_norm": 8.381549835205078,
      "learning_rate": 9.004389004389004e-06,
      "loss": 0.4058,
      "step": 913
    },
    {
      "epoch": 0.9501039501039501,
      "grad_norm": 13.460283279418945,
      "learning_rate": 9.002079002079002e-06,
      "loss": 0.6386,
      "step": 914
    },
    {
      "epoch": 0.9511434511434511,
      "grad_norm": 16.521923065185547,
      "learning_rate": 8.999768999769e-06,
      "loss": 0.9705,
      "step": 915
    },
    {
      "epoch": 0.9521829521829522,
      "grad_norm": 3.118180751800537,
      "learning_rate": 8.997458997458997e-06,
      "loss": 0.0535,
      "step": 916
    },
    {
      "epoch": 0.9532224532224532,
      "grad_norm": 3.885518789291382,
      "learning_rate": 8.995148995148995e-06,
      "loss": 0.0602,
      "step": 917
    },
    {
      "epoch": 0.9542619542619543,
      "grad_norm": 15.75216007232666,
      "learning_rate": 8.992838992838993e-06,
      "loss": 0.9788,
      "step": 918
    },
    {
      "epoch": 0.9553014553014553,
      "grad_norm": 15.924515724182129,
      "learning_rate": 8.990528990528992e-06,
      "loss": 0.7702,
      "step": 919
    },
    {
      "epoch": 0.9563409563409564,
      "grad_norm": 8.416720390319824,
      "learning_rate": 8.988218988218988e-06,
      "loss": 0.4784,
      "step": 920
    },
    {
      "epoch": 0.9573804573804574,
      "grad_norm": 9.229931831359863,
      "learning_rate": 8.985908985908986e-06,
      "loss": 1.115,
      "step": 921
    },
    {
      "epoch": 0.9584199584199584,
      "grad_norm": 4.722958564758301,
      "learning_rate": 8.983598983598985e-06,
      "loss": 0.1893,
      "step": 922
    },
    {
      "epoch": 0.9594594594594594,
      "grad_norm": 6.5108137130737305,
      "learning_rate": 8.981288981288983e-06,
      "loss": 0.1334,
      "step": 923
    },
    {
      "epoch": 0.9604989604989606,
      "grad_norm": 16.13971519470215,
      "learning_rate": 8.97897897897898e-06,
      "loss": 1.6763,
      "step": 924
    },
    {
      "epoch": 0.9615384615384616,
      "grad_norm": 6.114139556884766,
      "learning_rate": 8.976668976668977e-06,
      "loss": 0.1802,
      "step": 925
    },
    {
      "epoch": 0.9625779625779626,
      "grad_norm": 17.669862747192383,
      "learning_rate": 8.974358974358976e-06,
      "loss": 1.1319,
      "step": 926
    },
    {
      "epoch": 0.9636174636174636,
      "grad_norm": 2.548318386077881,
      "learning_rate": 8.972048972048972e-06,
      "loss": 0.1018,
      "step": 927
    },
    {
      "epoch": 0.9646569646569647,
      "grad_norm": 7.308948040008545,
      "learning_rate": 8.96973896973897e-06,
      "loss": 0.1264,
      "step": 928
    },
    {
      "epoch": 0.9656964656964657,
      "grad_norm": 3.920917272567749,
      "learning_rate": 8.967428967428968e-06,
      "loss": 0.1023,
      "step": 929
    },
    {
      "epoch": 0.9667359667359667,
      "grad_norm": 10.714740753173828,
      "learning_rate": 8.965118965118967e-06,
      "loss": 0.7717,
      "step": 930
    },
    {
      "epoch": 0.9677754677754677,
      "grad_norm": 7.647885322570801,
      "learning_rate": 8.962808962808963e-06,
      "loss": 0.2193,
      "step": 931
    },
    {
      "epoch": 0.9688149688149689,
      "grad_norm": 9.926279067993164,
      "learning_rate": 8.960498960498961e-06,
      "loss": 0.4121,
      "step": 932
    },
    {
      "epoch": 0.9698544698544699,
      "grad_norm": 1.8114266395568848,
      "learning_rate": 8.95818895818896e-06,
      "loss": 0.0401,
      "step": 933
    },
    {
      "epoch": 0.9708939708939709,
      "grad_norm": 13.62614917755127,
      "learning_rate": 8.955878955878956e-06,
      "loss": 0.8964,
      "step": 934
    },
    {
      "epoch": 0.9719334719334719,
      "grad_norm": 1.0622432231903076,
      "learning_rate": 8.953568953568954e-06,
      "loss": 0.0155,
      "step": 935
    },
    {
      "epoch": 0.972972972972973,
      "grad_norm": 6.726819038391113,
      "learning_rate": 8.951258951258952e-06,
      "loss": 0.1767,
      "step": 936
    },
    {
      "epoch": 0.974012474012474,
      "grad_norm": 8.454751014709473,
      "learning_rate": 8.94894894894895e-06,
      "loss": 0.5787,
      "step": 937
    },
    {
      "epoch": 0.975051975051975,
      "grad_norm": 1.420302152633667,
      "learning_rate": 8.946638946638947e-06,
      "loss": 0.0451,
      "step": 938
    },
    {
      "epoch": 0.9760914760914761,
      "grad_norm": 3.865201234817505,
      "learning_rate": 8.944328944328945e-06,
      "loss": 0.0869,
      "step": 939
    },
    {
      "epoch": 0.9771309771309772,
      "grad_norm": 4.933730125427246,
      "learning_rate": 8.942018942018943e-06,
      "loss": 0.0811,
      "step": 940
    },
    {
      "epoch": 0.9781704781704782,
      "grad_norm": 0.5709061622619629,
      "learning_rate": 8.939708939708942e-06,
      "loss": 0.0124,
      "step": 941
    },
    {
      "epoch": 0.9792099792099792,
      "grad_norm": 5.7213053703308105,
      "learning_rate": 8.937398937398938e-06,
      "loss": 0.0675,
      "step": 942
    },
    {
      "epoch": 0.9802494802494802,
      "grad_norm": 18.375486373901367,
      "learning_rate": 8.935088935088936e-06,
      "loss": 0.4712,
      "step": 943
    },
    {
      "epoch": 0.9812889812889813,
      "grad_norm": 12.987750053405762,
      "learning_rate": 8.932778932778935e-06,
      "loss": 0.2181,
      "step": 944
    },
    {
      "epoch": 0.9823284823284824,
      "grad_norm": 2.9199960231781006,
      "learning_rate": 8.930468930468931e-06,
      "loss": 0.0576,
      "step": 945
    },
    {
      "epoch": 0.9833679833679834,
      "grad_norm": 8.773855209350586,
      "learning_rate": 8.92815892815893e-06,
      "loss": 0.1334,
      "step": 946
    },
    {
      "epoch": 0.9844074844074844,
      "grad_norm": 3.0844333171844482,
      "learning_rate": 8.925848925848927e-06,
      "loss": 0.1197,
      "step": 947
    },
    {
      "epoch": 0.9854469854469855,
      "grad_norm": 7.7949066162109375,
      "learning_rate": 8.923538923538924e-06,
      "loss": 0.4242,
      "step": 948
    },
    {
      "epoch": 0.9864864864864865,
      "grad_norm": 3.214975595474243,
      "learning_rate": 8.921228921228922e-06,
      "loss": 0.0894,
      "step": 949
    },
    {
      "epoch": 0.9875259875259875,
      "grad_norm": 3.6018924713134766,
      "learning_rate": 8.91891891891892e-06,
      "loss": 0.2834,
      "step": 950
    },
    {
      "epoch": 0.9885654885654885,
      "grad_norm": 7.012938499450684,
      "learning_rate": 8.916608916608917e-06,
      "loss": 0.1426,
      "step": 951
    },
    {
      "epoch": 0.9896049896049897,
      "grad_norm": 23.49381446838379,
      "learning_rate": 8.914298914298915e-06,
      "loss": 1.7739,
      "step": 952
    },
    {
      "epoch": 0.9906444906444907,
      "grad_norm": 13.419476509094238,
      "learning_rate": 8.911988911988913e-06,
      "loss": 0.683,
      "step": 953
    },
    {
      "epoch": 0.9916839916839917,
      "grad_norm": 10.06177043914795,
      "learning_rate": 8.90967890967891e-06,
      "loss": 0.541,
      "step": 954
    },
    {
      "epoch": 0.9927234927234927,
      "grad_norm": 11.302132606506348,
      "learning_rate": 8.907368907368908e-06,
      "loss": 0.6544,
      "step": 955
    },
    {
      "epoch": 0.9937629937629938,
      "grad_norm": 2.3033945560455322,
      "learning_rate": 8.905058905058906e-06,
      "loss": 0.0506,
      "step": 956
    },
    {
      "epoch": 0.9948024948024948,
      "grad_norm": 8.643536567687988,
      "learning_rate": 8.902748902748903e-06,
      "loss": 0.4236,
      "step": 957
    },
    {
      "epoch": 0.9958419958419958,
      "grad_norm": 10.508223533630371,
      "learning_rate": 8.9004389004389e-06,
      "loss": 0.8041,
      "step": 958
    },
    {
      "epoch": 0.9968814968814969,
      "grad_norm": 16.5745792388916,
      "learning_rate": 8.898128898128899e-06,
      "loss": 0.7238,
      "step": 959
    },
    {
      "epoch": 0.997920997920998,
      "grad_norm": 6.697941303253174,
      "learning_rate": 8.895818895818895e-06,
      "loss": 0.1965,
      "step": 960
    },
    {
      "epoch": 0.998960498960499,
      "grad_norm": 1.8422704935073853,
      "learning_rate": 8.893508893508894e-06,
      "loss": 0.0437,
      "step": 961
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.4980247020721436,
      "learning_rate": 8.891198891198892e-06,
      "loss": 0.0669,
      "step": 962
    },
    {
      "epoch": 1.001039501039501,
      "grad_norm": 5.414978504180908,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.2941,
      "step": 963
    },
    {
      "epoch": 1.002079002079002,
      "grad_norm": 10.896177291870117,
      "learning_rate": 8.886578886578886e-06,
      "loss": 0.7354,
      "step": 964
    },
    {
      "epoch": 1.003118503118503,
      "grad_norm": 11.286123275756836,
      "learning_rate": 8.884268884268885e-06,
      "loss": 0.4795,
      "step": 965
    },
    {
      "epoch": 1.004158004158004,
      "grad_norm": 16.44017219543457,
      "learning_rate": 8.881958881958883e-06,
      "loss": 0.683,
      "step": 966
    },
    {
      "epoch": 1.0051975051975053,
      "grad_norm": 6.277804374694824,
      "learning_rate": 8.87964887964888e-06,
      "loss": 0.2146,
      "step": 967
    },
    {
      "epoch": 1.0062370062370063,
      "grad_norm": 14.160048484802246,
      "learning_rate": 8.877338877338878e-06,
      "loss": 0.9587,
      "step": 968
    },
    {
      "epoch": 1.0072765072765073,
      "grad_norm": 1.9493069648742676,
      "learning_rate": 8.875028875028876e-06,
      "loss": 0.0522,
      "step": 969
    },
    {
      "epoch": 1.0083160083160083,
      "grad_norm": 1.8083431720733643,
      "learning_rate": 8.872718872718874e-06,
      "loss": 0.0541,
      "step": 970
    },
    {
      "epoch": 1.0093555093555093,
      "grad_norm": 2.2266340255737305,
      "learning_rate": 8.87040887040887e-06,
      "loss": 0.033,
      "step": 971
    },
    {
      "epoch": 1.0103950103950103,
      "grad_norm": 0.941646933555603,
      "learning_rate": 8.868098868098869e-06,
      "loss": 0.0124,
      "step": 972
    },
    {
      "epoch": 1.0114345114345114,
      "grad_norm": 7.98224401473999,
      "learning_rate": 8.865788865788867e-06,
      "loss": 0.3622,
      "step": 973
    },
    {
      "epoch": 1.0124740124740126,
      "grad_norm": 5.987460613250732,
      "learning_rate": 8.863478863478863e-06,
      "loss": 0.2252,
      "step": 974
    },
    {
      "epoch": 1.0135135135135136,
      "grad_norm": 11.708882331848145,
      "learning_rate": 8.861168861168862e-06,
      "loss": 0.411,
      "step": 975
    },
    {
      "epoch": 1.0145530145530146,
      "grad_norm": 8.168597221374512,
      "learning_rate": 8.85885885885886e-06,
      "loss": 0.3952,
      "step": 976
    },
    {
      "epoch": 1.0155925155925156,
      "grad_norm": 0.640728235244751,
      "learning_rate": 8.856548856548858e-06,
      "loss": 0.015,
      "step": 977
    },
    {
      "epoch": 1.0166320166320166,
      "grad_norm": 10.76993465423584,
      "learning_rate": 8.854238854238854e-06,
      "loss": 0.4832,
      "step": 978
    },
    {
      "epoch": 1.0176715176715176,
      "grad_norm": 12.423575401306152,
      "learning_rate": 8.851928851928853e-06,
      "loss": 0.3292,
      "step": 979
    },
    {
      "epoch": 1.0187110187110187,
      "grad_norm": 13.478914260864258,
      "learning_rate": 8.84961884961885e-06,
      "loss": 0.8858,
      "step": 980
    },
    {
      "epoch": 1.0197505197505197,
      "grad_norm": 9.08768367767334,
      "learning_rate": 8.847308847308847e-06,
      "loss": 0.4073,
      "step": 981
    },
    {
      "epoch": 1.0207900207900207,
      "grad_norm": 0.12124160677194595,
      "learning_rate": 8.844998844998845e-06,
      "loss": 0.0027,
      "step": 982
    },
    {
      "epoch": 1.021829521829522,
      "grad_norm": 15.824604988098145,
      "learning_rate": 8.842688842688844e-06,
      "loss": 1.2646,
      "step": 983
    },
    {
      "epoch": 1.022869022869023,
      "grad_norm": 1.92180597782135,
      "learning_rate": 8.840378840378842e-06,
      "loss": 0.0305,
      "step": 984
    },
    {
      "epoch": 1.023908523908524,
      "grad_norm": 3.483933210372925,
      "learning_rate": 8.838068838068838e-06,
      "loss": 0.0649,
      "step": 985
    },
    {
      "epoch": 1.024948024948025,
      "grad_norm": 7.1540045738220215,
      "learning_rate": 8.835758835758837e-06,
      "loss": 0.3779,
      "step": 986
    },
    {
      "epoch": 1.025987525987526,
      "grad_norm": 0.35930129885673523,
      "learning_rate": 8.833448833448835e-06,
      "loss": 0.0067,
      "step": 987
    },
    {
      "epoch": 1.027027027027027,
      "grad_norm": 5.606091022491455,
      "learning_rate": 8.831138831138833e-06,
      "loss": 0.1168,
      "step": 988
    },
    {
      "epoch": 1.028066528066528,
      "grad_norm": 13.382498741149902,
      "learning_rate": 8.82882882882883e-06,
      "loss": 1.0233,
      "step": 989
    },
    {
      "epoch": 1.0291060291060292,
      "grad_norm": 8.95033073425293,
      "learning_rate": 8.826518826518828e-06,
      "loss": 1.1854,
      "step": 990
    },
    {
      "epoch": 1.0301455301455302,
      "grad_norm": 10.991436004638672,
      "learning_rate": 8.824208824208826e-06,
      "loss": 0.8696,
      "step": 991
    },
    {
      "epoch": 1.0311850311850312,
      "grad_norm": 17.413393020629883,
      "learning_rate": 8.821898821898822e-06,
      "loss": 0.3927,
      "step": 992
    },
    {
      "epoch": 1.0322245322245323,
      "grad_norm": 0.3354548513889313,
      "learning_rate": 8.81958881958882e-06,
      "loss": 0.0095,
      "step": 993
    },
    {
      "epoch": 1.0332640332640333,
      "grad_norm": 3.1128478050231934,
      "learning_rate": 8.817278817278819e-06,
      "loss": 0.1201,
      "step": 994
    },
    {
      "epoch": 1.0343035343035343,
      "grad_norm": 1.8844947814941406,
      "learning_rate": 8.814968814968817e-06,
      "loss": 0.0735,
      "step": 995
    },
    {
      "epoch": 1.0353430353430353,
      "grad_norm": 3.381255626678467,
      "learning_rate": 8.812658812658813e-06,
      "loss": 0.0986,
      "step": 996
    },
    {
      "epoch": 1.0363825363825363,
      "grad_norm": 11.728659629821777,
      "learning_rate": 8.810348810348812e-06,
      "loss": 0.3749,
      "step": 997
    },
    {
      "epoch": 1.0374220374220373,
      "grad_norm": 9.866026878356934,
      "learning_rate": 8.80803880803881e-06,
      "loss": 0.4012,
      "step": 998
    },
    {
      "epoch": 1.0384615384615385,
      "grad_norm": 7.215566635131836,
      "learning_rate": 8.805728805728806e-06,
      "loss": 0.2596,
      "step": 999
    },
    {
      "epoch": 1.0395010395010396,
      "grad_norm": 0.8678320646286011,
      "learning_rate": 8.803418803418804e-06,
      "loss": 0.0259,
      "step": 1000
    },
    {
      "epoch": 1.0405405405405406,
      "grad_norm": 2.84258770942688,
      "learning_rate": 8.801108801108803e-06,
      "loss": 0.0602,
      "step": 1001
    },
    {
      "epoch": 1.0415800415800416,
      "grad_norm": 0.3957040309906006,
      "learning_rate": 8.798798798798799e-06,
      "loss": 0.006,
      "step": 1002
    },
    {
      "epoch": 1.0426195426195426,
      "grad_norm": 10.925928115844727,
      "learning_rate": 8.796488796488797e-06,
      "loss": 0.7463,
      "step": 1003
    },
    {
      "epoch": 1.0436590436590436,
      "grad_norm": 9.018205642700195,
      "learning_rate": 8.794178794178795e-06,
      "loss": 0.826,
      "step": 1004
    },
    {
      "epoch": 1.0446985446985446,
      "grad_norm": 1.216611623764038,
      "learning_rate": 8.791868791868794e-06,
      "loss": 0.0239,
      "step": 1005
    },
    {
      "epoch": 1.0457380457380459,
      "grad_norm": 0.12714122235774994,
      "learning_rate": 8.78955878955879e-06,
      "loss": 0.0023,
      "step": 1006
    },
    {
      "epoch": 1.0467775467775469,
      "grad_norm": 1.0121233463287354,
      "learning_rate": 8.787248787248788e-06,
      "loss": 0.0178,
      "step": 1007
    },
    {
      "epoch": 1.0478170478170479,
      "grad_norm": 5.475372314453125,
      "learning_rate": 8.784938784938787e-06,
      "loss": 0.0919,
      "step": 1008
    },
    {
      "epoch": 1.0488565488565489,
      "grad_norm": 2.838118553161621,
      "learning_rate": 8.782628782628783e-06,
      "loss": 0.0921,
      "step": 1009
    },
    {
      "epoch": 1.04989604989605,
      "grad_norm": 9.721406936645508,
      "learning_rate": 8.780318780318781e-06,
      "loss": 0.3828,
      "step": 1010
    },
    {
      "epoch": 1.050935550935551,
      "grad_norm": 9.811200141906738,
      "learning_rate": 8.77800877800878e-06,
      "loss": 0.5693,
      "step": 1011
    },
    {
      "epoch": 1.051975051975052,
      "grad_norm": 1.1226965188980103,
      "learning_rate": 8.775698775698776e-06,
      "loss": 0.0177,
      "step": 1012
    },
    {
      "epoch": 1.053014553014553,
      "grad_norm": 2.9267220497131348,
      "learning_rate": 8.773388773388774e-06,
      "loss": 0.0449,
      "step": 1013
    },
    {
      "epoch": 1.054054054054054,
      "grad_norm": 9.672924041748047,
      "learning_rate": 8.771078771078772e-06,
      "loss": 0.2402,
      "step": 1014
    },
    {
      "epoch": 1.0550935550935552,
      "grad_norm": 11.029509544372559,
      "learning_rate": 8.768768768768769e-06,
      "loss": 0.5824,
      "step": 1015
    },
    {
      "epoch": 1.0561330561330562,
      "grad_norm": 16.500804901123047,
      "learning_rate": 8.766458766458767e-06,
      "loss": 1.3339,
      "step": 1016
    },
    {
      "epoch": 1.0571725571725572,
      "grad_norm": 3.943711757659912,
      "learning_rate": 8.764148764148765e-06,
      "loss": 0.1545,
      "step": 1017
    },
    {
      "epoch": 1.0582120582120582,
      "grad_norm": 8.231452941894531,
      "learning_rate": 8.761838761838762e-06,
      "loss": 0.3989,
      "step": 1018
    },
    {
      "epoch": 1.0592515592515592,
      "grad_norm": 7.401911735534668,
      "learning_rate": 8.75952875952876e-06,
      "loss": 0.1997,
      "step": 1019
    },
    {
      "epoch": 1.0602910602910602,
      "grad_norm": 12.006552696228027,
      "learning_rate": 8.757218757218758e-06,
      "loss": 0.6828,
      "step": 1020
    },
    {
      "epoch": 1.0613305613305613,
      "grad_norm": 24.606016159057617,
      "learning_rate": 8.754908754908755e-06,
      "loss": 1.5875,
      "step": 1021
    },
    {
      "epoch": 1.0623700623700625,
      "grad_norm": 18.473796844482422,
      "learning_rate": 8.752598752598753e-06,
      "loss": 0.8307,
      "step": 1022
    },
    {
      "epoch": 1.0634095634095635,
      "grad_norm": 21.40069580078125,
      "learning_rate": 8.750288750288751e-06,
      "loss": 0.5301,
      "step": 1023
    },
    {
      "epoch": 1.0644490644490645,
      "grad_norm": 8.655167579650879,
      "learning_rate": 8.747978747978749e-06,
      "loss": 0.3022,
      "step": 1024
    },
    {
      "epoch": 1.0654885654885655,
      "grad_norm": 13.602567672729492,
      "learning_rate": 8.745668745668746e-06,
      "loss": 0.373,
      "step": 1025
    },
    {
      "epoch": 1.0665280665280665,
      "grad_norm": 2.638014793395996,
      "learning_rate": 8.743358743358744e-06,
      "loss": 0.0676,
      "step": 1026
    },
    {
      "epoch": 1.0675675675675675,
      "grad_norm": 7.661709308624268,
      "learning_rate": 8.741048741048742e-06,
      "loss": 0.2338,
      "step": 1027
    },
    {
      "epoch": 1.0686070686070686,
      "grad_norm": 7.508708477020264,
      "learning_rate": 8.738738738738739e-06,
      "loss": 0.4082,
      "step": 1028
    },
    {
      "epoch": 1.0696465696465696,
      "grad_norm": 3.627363443374634,
      "learning_rate": 8.736428736428737e-06,
      "loss": 0.0684,
      "step": 1029
    },
    {
      "epoch": 1.0706860706860706,
      "grad_norm": 5.621256351470947,
      "learning_rate": 8.734118734118735e-06,
      "loss": 0.1047,
      "step": 1030
    },
    {
      "epoch": 1.0717255717255718,
      "grad_norm": 0.027766048908233643,
      "learning_rate": 8.731808731808733e-06,
      "loss": 0.0008,
      "step": 1031
    },
    {
      "epoch": 1.0727650727650728,
      "grad_norm": 0.6192247867584229,
      "learning_rate": 8.72949872949873e-06,
      "loss": 0.0086,
      "step": 1032
    },
    {
      "epoch": 1.0738045738045738,
      "grad_norm": 5.98634672164917,
      "learning_rate": 8.727188727188728e-06,
      "loss": 0.1553,
      "step": 1033
    },
    {
      "epoch": 1.0748440748440748,
      "grad_norm": 7.63029146194458,
      "learning_rate": 8.724878724878726e-06,
      "loss": 0.4071,
      "step": 1034
    },
    {
      "epoch": 1.0758835758835759,
      "grad_norm": 17.183271408081055,
      "learning_rate": 8.722568722568722e-06,
      "loss": 0.724,
      "step": 1035
    },
    {
      "epoch": 1.0769230769230769,
      "grad_norm": 8.9673490524292,
      "learning_rate": 8.72025872025872e-06,
      "loss": 0.0969,
      "step": 1036
    },
    {
      "epoch": 1.0779625779625779,
      "grad_norm": 9.389533996582031,
      "learning_rate": 8.717948717948719e-06,
      "loss": 0.7216,
      "step": 1037
    },
    {
      "epoch": 1.0790020790020791,
      "grad_norm": 7.645777702331543,
      "learning_rate": 8.715638715638717e-06,
      "loss": 0.3217,
      "step": 1038
    },
    {
      "epoch": 1.0800415800415801,
      "grad_norm": 1.5751149654388428,
      "learning_rate": 8.713328713328714e-06,
      "loss": 0.031,
      "step": 1039
    },
    {
      "epoch": 1.0810810810810811,
      "grad_norm": 2.261518955230713,
      "learning_rate": 8.711018711018712e-06,
      "loss": 0.0895,
      "step": 1040
    },
    {
      "epoch": 1.0821205821205822,
      "grad_norm": 0.32005324959754944,
      "learning_rate": 8.70870870870871e-06,
      "loss": 0.0096,
      "step": 1041
    },
    {
      "epoch": 1.0831600831600832,
      "grad_norm": 0.7632772922515869,
      "learning_rate": 8.706398706398708e-06,
      "loss": 0.0126,
      "step": 1042
    },
    {
      "epoch": 1.0841995841995842,
      "grad_norm": 3.8408350944519043,
      "learning_rate": 8.704088704088705e-06,
      "loss": 0.0494,
      "step": 1043
    },
    {
      "epoch": 1.0852390852390852,
      "grad_norm": 6.506851673126221,
      "learning_rate": 8.701778701778703e-06,
      "loss": 0.1149,
      "step": 1044
    },
    {
      "epoch": 1.0862785862785862,
      "grad_norm": 5.510050296783447,
      "learning_rate": 8.699468699468701e-06,
      "loss": 0.1562,
      "step": 1045
    },
    {
      "epoch": 1.0873180873180872,
      "grad_norm": 0.16971538960933685,
      "learning_rate": 8.697158697158697e-06,
      "loss": 0.003,
      "step": 1046
    },
    {
      "epoch": 1.0883575883575884,
      "grad_norm": 9.97668170928955,
      "learning_rate": 8.694848694848696e-06,
      "loss": 0.6033,
      "step": 1047
    },
    {
      "epoch": 1.0893970893970895,
      "grad_norm": 0.341886967420578,
      "learning_rate": 8.692538692538694e-06,
      "loss": 0.0061,
      "step": 1048
    },
    {
      "epoch": 1.0904365904365905,
      "grad_norm": 10.055686950683594,
      "learning_rate": 8.690228690228692e-06,
      "loss": 0.2323,
      "step": 1049
    },
    {
      "epoch": 1.0914760914760915,
      "grad_norm": 6.149728775024414,
      "learning_rate": 8.687918687918689e-06,
      "loss": 0.2168,
      "step": 1050
    },
    {
      "epoch": 1.0925155925155925,
      "grad_norm": 0.03411586210131645,
      "learning_rate": 8.685608685608687e-06,
      "loss": 0.0009,
      "step": 1051
    },
    {
      "epoch": 1.0935550935550935,
      "grad_norm": 1.4763023853302002,
      "learning_rate": 8.683298683298685e-06,
      "loss": 0.0403,
      "step": 1052
    },
    {
      "epoch": 1.0945945945945945,
      "grad_norm": 0.5744214653968811,
      "learning_rate": 8.680988680988681e-06,
      "loss": 0.0182,
      "step": 1053
    },
    {
      "epoch": 1.0956340956340958,
      "grad_norm": 8.413043975830078,
      "learning_rate": 8.67867867867868e-06,
      "loss": 0.4508,
      "step": 1054
    },
    {
      "epoch": 1.0966735966735968,
      "grad_norm": 3.0033912658691406,
      "learning_rate": 8.676368676368678e-06,
      "loss": 0.0878,
      "step": 1055
    },
    {
      "epoch": 1.0977130977130978,
      "grad_norm": 2.9098126888275146,
      "learning_rate": 8.674058674058676e-06,
      "loss": 0.123,
      "step": 1056
    },
    {
      "epoch": 1.0987525987525988,
      "grad_norm": 10.180469512939453,
      "learning_rate": 8.671748671748672e-06,
      "loss": 0.3434,
      "step": 1057
    },
    {
      "epoch": 1.0997920997920998,
      "grad_norm": 0.4921611249446869,
      "learning_rate": 8.66943866943867e-06,
      "loss": 0.0136,
      "step": 1058
    },
    {
      "epoch": 1.1008316008316008,
      "grad_norm": 13.588961601257324,
      "learning_rate": 8.667128667128669e-06,
      "loss": 1.4453,
      "step": 1059
    },
    {
      "epoch": 1.1018711018711018,
      "grad_norm": 14.660667419433594,
      "learning_rate": 8.664818664818665e-06,
      "loss": 0.5864,
      "step": 1060
    },
    {
      "epoch": 1.1029106029106028,
      "grad_norm": 5.017448902130127,
      "learning_rate": 8.662508662508664e-06,
      "loss": 0.1624,
      "step": 1061
    },
    {
      "epoch": 1.1039501039501038,
      "grad_norm": 17.048917770385742,
      "learning_rate": 8.660198660198662e-06,
      "loss": 1.5054,
      "step": 1062
    },
    {
      "epoch": 1.104989604989605,
      "grad_norm": 8.416142463684082,
      "learning_rate": 8.657888657888658e-06,
      "loss": 0.4395,
      "step": 1063
    },
    {
      "epoch": 1.106029106029106,
      "grad_norm": 10.862117767333984,
      "learning_rate": 8.655578655578656e-06,
      "loss": 0.9302,
      "step": 1064
    },
    {
      "epoch": 1.107068607068607,
      "grad_norm": 10.158036231994629,
      "learning_rate": 8.653268653268655e-06,
      "loss": 0.5861,
      "step": 1065
    },
    {
      "epoch": 1.1081081081081081,
      "grad_norm": 6.491422653198242,
      "learning_rate": 8.650958650958651e-06,
      "loss": 0.1169,
      "step": 1066
    },
    {
      "epoch": 1.1091476091476091,
      "grad_norm": 6.694852352142334,
      "learning_rate": 8.64864864864865e-06,
      "loss": 0.1717,
      "step": 1067
    },
    {
      "epoch": 1.1101871101871101,
      "grad_norm": 5.018531799316406,
      "learning_rate": 8.646338646338648e-06,
      "loss": 0.1341,
      "step": 1068
    },
    {
      "epoch": 1.1112266112266111,
      "grad_norm": 1.5225193500518799,
      "learning_rate": 8.644028644028644e-06,
      "loss": 0.0356,
      "step": 1069
    },
    {
      "epoch": 1.1122661122661124,
      "grad_norm": 2.1500790119171143,
      "learning_rate": 8.641718641718642e-06,
      "loss": 0.0403,
      "step": 1070
    },
    {
      "epoch": 1.1133056133056134,
      "grad_norm": 33.20718002319336,
      "learning_rate": 8.63940863940864e-06,
      "loss": 0.1609,
      "step": 1071
    },
    {
      "epoch": 1.1143451143451144,
      "grad_norm": 16.251794815063477,
      "learning_rate": 8.637098637098637e-06,
      "loss": 0.5655,
      "step": 1072
    },
    {
      "epoch": 1.1153846153846154,
      "grad_norm": 3.3469290733337402,
      "learning_rate": 8.634788634788635e-06,
      "loss": 0.0595,
      "step": 1073
    },
    {
      "epoch": 1.1164241164241164,
      "grad_norm": 0.12442325055599213,
      "learning_rate": 8.632478632478633e-06,
      "loss": 0.002,
      "step": 1074
    },
    {
      "epoch": 1.1174636174636174,
      "grad_norm": 0.4608583450317383,
      "learning_rate": 8.63016863016863e-06,
      "loss": 0.0039,
      "step": 1075
    },
    {
      "epoch": 1.1185031185031185,
      "grad_norm": 6.6315717697143555,
      "learning_rate": 8.627858627858628e-06,
      "loss": 0.1658,
      "step": 1076
    },
    {
      "epoch": 1.1195426195426195,
      "grad_norm": 14.120782852172852,
      "learning_rate": 8.625548625548626e-06,
      "loss": 1.1364,
      "step": 1077
    },
    {
      "epoch": 1.1205821205821205,
      "grad_norm": 12.094304084777832,
      "learning_rate": 8.623238623238624e-06,
      "loss": 1.4489,
      "step": 1078
    },
    {
      "epoch": 1.1216216216216217,
      "grad_norm": 14.251092910766602,
      "learning_rate": 8.62092862092862e-06,
      "loss": 1.493,
      "step": 1079
    },
    {
      "epoch": 1.1226611226611227,
      "grad_norm": 2.4907310009002686,
      "learning_rate": 8.618618618618619e-06,
      "loss": 0.0375,
      "step": 1080
    },
    {
      "epoch": 1.1237006237006237,
      "grad_norm": 10.427132606506348,
      "learning_rate": 8.616308616308617e-06,
      "loss": 0.2393,
      "step": 1081
    },
    {
      "epoch": 1.1247401247401247,
      "grad_norm": 0.9016990661621094,
      "learning_rate": 8.613998613998614e-06,
      "loss": 0.024,
      "step": 1082
    },
    {
      "epoch": 1.1257796257796258,
      "grad_norm": 13.01040267944336,
      "learning_rate": 8.611688611688612e-06,
      "loss": 0.3459,
      "step": 1083
    },
    {
      "epoch": 1.1268191268191268,
      "grad_norm": 0.26096540689468384,
      "learning_rate": 8.60937860937861e-06,
      "loss": 0.0075,
      "step": 1084
    },
    {
      "epoch": 1.1278586278586278,
      "grad_norm": 5.673876762390137,
      "learning_rate": 8.607068607068608e-06,
      "loss": 0.2176,
      "step": 1085
    },
    {
      "epoch": 1.128898128898129,
      "grad_norm": 1.2679526805877686,
      "learning_rate": 8.604758604758605e-06,
      "loss": 0.0507,
      "step": 1086
    },
    {
      "epoch": 1.12993762993763,
      "grad_norm": 10.18384838104248,
      "learning_rate": 8.602448602448603e-06,
      "loss": 0.9842,
      "step": 1087
    },
    {
      "epoch": 1.130977130977131,
      "grad_norm": 1.5802457332611084,
      "learning_rate": 8.600138600138601e-06,
      "loss": 0.0257,
      "step": 1088
    },
    {
      "epoch": 1.132016632016632,
      "grad_norm": 11.541592597961426,
      "learning_rate": 8.597828597828598e-06,
      "loss": 1.3012,
      "step": 1089
    },
    {
      "epoch": 1.133056133056133,
      "grad_norm": 9.967923164367676,
      "learning_rate": 8.595518595518596e-06,
      "loss": 0.8518,
      "step": 1090
    },
    {
      "epoch": 1.134095634095634,
      "grad_norm": 1.6320937871932983,
      "learning_rate": 8.593208593208594e-06,
      "loss": 0.0294,
      "step": 1091
    },
    {
      "epoch": 1.135135135135135,
      "grad_norm": 18.003585815429688,
      "learning_rate": 8.590898590898592e-06,
      "loss": 1.2795,
      "step": 1092
    },
    {
      "epoch": 1.136174636174636,
      "grad_norm": 9.14699935913086,
      "learning_rate": 8.588588588588589e-06,
      "loss": 0.2811,
      "step": 1093
    },
    {
      "epoch": 1.137214137214137,
      "grad_norm": 8.81598949432373,
      "learning_rate": 8.586278586278587e-06,
      "loss": 0.4506,
      "step": 1094
    },
    {
      "epoch": 1.1382536382536383,
      "grad_norm": 10.513948440551758,
      "learning_rate": 8.583968583968585e-06,
      "loss": 0.2852,
      "step": 1095
    },
    {
      "epoch": 1.1392931392931394,
      "grad_norm": 10.706802368164062,
      "learning_rate": 8.581658581658583e-06,
      "loss": 0.3514,
      "step": 1096
    },
    {
      "epoch": 1.1403326403326404,
      "grad_norm": 9.5643310546875,
      "learning_rate": 8.57934857934858e-06,
      "loss": 1.0022,
      "step": 1097
    },
    {
      "epoch": 1.1413721413721414,
      "grad_norm": 3.270961284637451,
      "learning_rate": 8.577038577038578e-06,
      "loss": 0.0555,
      "step": 1098
    },
    {
      "epoch": 1.1424116424116424,
      "grad_norm": 11.908968925476074,
      "learning_rate": 8.574728574728576e-06,
      "loss": 0.2819,
      "step": 1099
    },
    {
      "epoch": 1.1434511434511434,
      "grad_norm": 9.292601585388184,
      "learning_rate": 8.572418572418573e-06,
      "loss": 0.3581,
      "step": 1100
    },
    {
      "epoch": 1.1444906444906444,
      "grad_norm": 0.7729955911636353,
      "learning_rate": 8.570108570108571e-06,
      "loss": 0.0193,
      "step": 1101
    },
    {
      "epoch": 1.1455301455301456,
      "grad_norm": 1.3490409851074219,
      "learning_rate": 8.567798567798569e-06,
      "loss": 0.0295,
      "step": 1102
    },
    {
      "epoch": 1.1465696465696467,
      "grad_norm": 9.878353118896484,
      "learning_rate": 8.565488565488567e-06,
      "loss": 0.4995,
      "step": 1103
    },
    {
      "epoch": 1.1476091476091477,
      "grad_norm": 8.845831871032715,
      "learning_rate": 8.563178563178564e-06,
      "loss": 0.6224,
      "step": 1104
    },
    {
      "epoch": 1.1486486486486487,
      "grad_norm": 0.5114465951919556,
      "learning_rate": 8.560868560868562e-06,
      "loss": 0.0118,
      "step": 1105
    },
    {
      "epoch": 1.1496881496881497,
      "grad_norm": 0.6183711290359497,
      "learning_rate": 8.55855855855856e-06,
      "loss": 0.0184,
      "step": 1106
    },
    {
      "epoch": 1.1507276507276507,
      "grad_norm": 0.25186434388160706,
      "learning_rate": 8.556248556248557e-06,
      "loss": 0.0053,
      "step": 1107
    },
    {
      "epoch": 1.1517671517671517,
      "grad_norm": 2.6005747318267822,
      "learning_rate": 8.553938553938555e-06,
      "loss": 0.0756,
      "step": 1108
    },
    {
      "epoch": 1.1528066528066527,
      "grad_norm": 6.125280380249023,
      "learning_rate": 8.551628551628553e-06,
      "loss": 0.3256,
      "step": 1109
    },
    {
      "epoch": 1.1538461538461537,
      "grad_norm": 6.985195636749268,
      "learning_rate": 8.549318549318551e-06,
      "loss": 0.3697,
      "step": 1110
    },
    {
      "epoch": 1.154885654885655,
      "grad_norm": 7.720252513885498,
      "learning_rate": 8.547008547008548e-06,
      "loss": 0.3452,
      "step": 1111
    },
    {
      "epoch": 1.155925155925156,
      "grad_norm": 0.41022229194641113,
      "learning_rate": 8.544698544698546e-06,
      "loss": 0.0086,
      "step": 1112
    },
    {
      "epoch": 1.156964656964657,
      "grad_norm": 4.509541988372803,
      "learning_rate": 8.542388542388544e-06,
      "loss": 0.1699,
      "step": 1113
    },
    {
      "epoch": 1.158004158004158,
      "grad_norm": 0.026520010083913803,
      "learning_rate": 8.54007854007854e-06,
      "loss": 0.0009,
      "step": 1114
    },
    {
      "epoch": 1.159043659043659,
      "grad_norm": 1.7383307218551636,
      "learning_rate": 8.537768537768539e-06,
      "loss": 0.0466,
      "step": 1115
    },
    {
      "epoch": 1.16008316008316,
      "grad_norm": 5.903339385986328,
      "learning_rate": 8.535458535458537e-06,
      "loss": 0.2287,
      "step": 1116
    },
    {
      "epoch": 1.161122661122661,
      "grad_norm": 2.3729774951934814,
      "learning_rate": 8.533148533148533e-06,
      "loss": 0.0557,
      "step": 1117
    },
    {
      "epoch": 1.1621621621621623,
      "grad_norm": 7.9382548332214355,
      "learning_rate": 8.530838530838532e-06,
      "loss": 0.3049,
      "step": 1118
    },
    {
      "epoch": 1.1632016632016633,
      "grad_norm": 19.727888107299805,
      "learning_rate": 8.52852852852853e-06,
      "loss": 1.9244,
      "step": 1119
    },
    {
      "epoch": 1.1642411642411643,
      "grad_norm": 8.529170989990234,
      "learning_rate": 8.526218526218526e-06,
      "loss": 0.4675,
      "step": 1120
    },
    {
      "epoch": 1.1652806652806653,
      "grad_norm": 3.230149507522583,
      "learning_rate": 8.523908523908525e-06,
      "loss": 0.0742,
      "step": 1121
    },
    {
      "epoch": 1.1663201663201663,
      "grad_norm": 2.254007577896118,
      "learning_rate": 8.521598521598523e-06,
      "loss": 0.0609,
      "step": 1122
    },
    {
      "epoch": 1.1673596673596673,
      "grad_norm": 7.165493011474609,
      "learning_rate": 8.51928851928852e-06,
      "loss": 0.4766,
      "step": 1123
    },
    {
      "epoch": 1.1683991683991684,
      "grad_norm": 6.5497145652771,
      "learning_rate": 8.516978516978517e-06,
      "loss": 0.1697,
      "step": 1124
    },
    {
      "epoch": 1.1694386694386694,
      "grad_norm": 7.064365863800049,
      "learning_rate": 8.514668514668516e-06,
      "loss": 0.1954,
      "step": 1125
    },
    {
      "epoch": 1.1704781704781704,
      "grad_norm": 1.0393562316894531,
      "learning_rate": 8.512358512358512e-06,
      "loss": 0.029,
      "step": 1126
    },
    {
      "epoch": 1.1715176715176716,
      "grad_norm": 14.624014854431152,
      "learning_rate": 8.51004851004851e-06,
      "loss": 0.3145,
      "step": 1127
    },
    {
      "epoch": 1.1725571725571726,
      "grad_norm": 11.193547248840332,
      "learning_rate": 8.507738507738508e-06,
      "loss": 0.785,
      "step": 1128
    },
    {
      "epoch": 1.1735966735966736,
      "grad_norm": 9.468733787536621,
      "learning_rate": 8.505428505428505e-06,
      "loss": 0.5649,
      "step": 1129
    },
    {
      "epoch": 1.1746361746361746,
      "grad_norm": 14.636096954345703,
      "learning_rate": 8.503118503118503e-06,
      "loss": 1.0006,
      "step": 1130
    },
    {
      "epoch": 1.1756756756756757,
      "grad_norm": 0.3806226849555969,
      "learning_rate": 8.500808500808501e-06,
      "loss": 0.0093,
      "step": 1131
    },
    {
      "epoch": 1.1767151767151767,
      "grad_norm": 0.13574692606925964,
      "learning_rate": 8.4984984984985e-06,
      "loss": 0.0033,
      "step": 1132
    },
    {
      "epoch": 1.1777546777546777,
      "grad_norm": 8.421379089355469,
      "learning_rate": 8.496188496188496e-06,
      "loss": 0.2561,
      "step": 1133
    },
    {
      "epoch": 1.178794178794179,
      "grad_norm": 4.2164201736450195,
      "learning_rate": 8.493878493878494e-06,
      "loss": 0.2455,
      "step": 1134
    },
    {
      "epoch": 1.17983367983368,
      "grad_norm": 6.926000595092773,
      "learning_rate": 8.491568491568492e-06,
      "loss": 0.3277,
      "step": 1135
    },
    {
      "epoch": 1.180873180873181,
      "grad_norm": 3.1254656314849854,
      "learning_rate": 8.489258489258489e-06,
      "loss": 0.0762,
      "step": 1136
    },
    {
      "epoch": 1.181912681912682,
      "grad_norm": 0.08411677926778793,
      "learning_rate": 8.486948486948487e-06,
      "loss": 0.0014,
      "step": 1137
    },
    {
      "epoch": 1.182952182952183,
      "grad_norm": 1.5572375059127808,
      "learning_rate": 8.484638484638485e-06,
      "loss": 0.0215,
      "step": 1138
    },
    {
      "epoch": 1.183991683991684,
      "grad_norm": 1.5848573446273804,
      "learning_rate": 8.482328482328483e-06,
      "loss": 0.0225,
      "step": 1139
    },
    {
      "epoch": 1.185031185031185,
      "grad_norm": 1.0582151412963867,
      "learning_rate": 8.48001848001848e-06,
      "loss": 0.2594,
      "step": 1140
    },
    {
      "epoch": 1.186070686070686,
      "grad_norm": 9.030559539794922,
      "learning_rate": 8.477708477708478e-06,
      "loss": 0.2851,
      "step": 1141
    },
    {
      "epoch": 1.187110187110187,
      "grad_norm": 15.570144653320312,
      "learning_rate": 8.475398475398476e-06,
      "loss": 0.3463,
      "step": 1142
    },
    {
      "epoch": 1.1881496881496882,
      "grad_norm": 10.316506385803223,
      "learning_rate": 8.473088473088473e-06,
      "loss": 0.5758,
      "step": 1143
    },
    {
      "epoch": 1.1891891891891893,
      "grad_norm": 5.263421058654785,
      "learning_rate": 8.470778470778471e-06,
      "loss": 0.1077,
      "step": 1144
    },
    {
      "epoch": 1.1902286902286903,
      "grad_norm": 4.201646327972412,
      "learning_rate": 8.46846846846847e-06,
      "loss": 0.1171,
      "step": 1145
    },
    {
      "epoch": 1.1912681912681913,
      "grad_norm": 0.2046760469675064,
      "learning_rate": 8.466158466158467e-06,
      "loss": 0.0044,
      "step": 1146
    },
    {
      "epoch": 1.1923076923076923,
      "grad_norm": 2.979363203048706,
      "learning_rate": 8.463848463848464e-06,
      "loss": 0.0425,
      "step": 1147
    },
    {
      "epoch": 1.1933471933471933,
      "grad_norm": 18.337308883666992,
      "learning_rate": 8.461538461538462e-06,
      "loss": 1.5175,
      "step": 1148
    },
    {
      "epoch": 1.1943866943866943,
      "grad_norm": 3.2387545108795166,
      "learning_rate": 8.45922845922846e-06,
      "loss": 0.0629,
      "step": 1149
    },
    {
      "epoch": 1.1954261954261955,
      "grad_norm": 10.86263656616211,
      "learning_rate": 8.456918456918458e-06,
      "loss": 0.3675,
      "step": 1150
    },
    {
      "epoch": 1.1964656964656966,
      "grad_norm": NaN,
      "learning_rate": 8.454608454608455e-06,
      "loss": 0.0022,
      "step": 1151
    },
    {
      "epoch": 1.1975051975051976,
      "grad_norm": 1.9744138717651367,
      "learning_rate": 8.452298452298453e-06,
      "loss": 0.0545,
      "step": 1152
    },
    {
      "epoch": 1.1985446985446986,
      "grad_norm": 9.182241439819336,
      "learning_rate": 8.449988449988451e-06,
      "loss": 0.1764,
      "step": 1153
    },
    {
      "epoch": 1.1995841995841996,
      "grad_norm": 14.103419303894043,
      "learning_rate": 8.447678447678448e-06,
      "loss": 1.7113,
      "step": 1154
    },
    {
      "epoch": 1.2006237006237006,
      "grad_norm": 5.250944137573242,
      "learning_rate": 8.445368445368446e-06,
      "loss": 0.2917,
      "step": 1155
    },
    {
      "epoch": 1.2016632016632016,
      "grad_norm": 8.00632095336914,
      "learning_rate": 8.443058443058444e-06,
      "loss": 0.3142,
      "step": 1156
    },
    {
      "epoch": 1.2027027027027026,
      "grad_norm": 9.843782424926758,
      "learning_rate": 8.440748440748442e-06,
      "loss": 0.524,
      "step": 1157
    },
    {
      "epoch": 1.2037422037422036,
      "grad_norm": 0.603416919708252,
      "learning_rate": 8.438438438438439e-06,
      "loss": 0.0099,
      "step": 1158
    },
    {
      "epoch": 1.2047817047817049,
      "grad_norm": 8.074755668640137,
      "learning_rate": 8.436128436128437e-06,
      "loss": 0.2532,
      "step": 1159
    },
    {
      "epoch": 1.2058212058212059,
      "grad_norm": 8.356449127197266,
      "learning_rate": 8.433818433818435e-06,
      "loss": 1.1221,
      "step": 1160
    },
    {
      "epoch": 1.206860706860707,
      "grad_norm": 12.065170288085938,
      "learning_rate": 8.431508431508432e-06,
      "loss": 0.6152,
      "step": 1161
    },
    {
      "epoch": 1.207900207900208,
      "grad_norm": 1.928618311882019,
      "learning_rate": 8.42919842919843e-06,
      "loss": 0.039,
      "step": 1162
    },
    {
      "epoch": 1.208939708939709,
      "grad_norm": 3.5592713356018066,
      "learning_rate": 8.426888426888428e-06,
      "loss": 0.1305,
      "step": 1163
    },
    {
      "epoch": 1.20997920997921,
      "grad_norm": 6.7794671058654785,
      "learning_rate": 8.424578424578426e-06,
      "loss": 0.5323,
      "step": 1164
    },
    {
      "epoch": 1.211018711018711,
      "grad_norm": 10.264362335205078,
      "learning_rate": 8.422268422268423e-06,
      "loss": 0.3864,
      "step": 1165
    },
    {
      "epoch": 1.2120582120582122,
      "grad_norm": 11.034218788146973,
      "learning_rate": 8.419958419958421e-06,
      "loss": 0.3514,
      "step": 1166
    },
    {
      "epoch": 1.2130977130977132,
      "grad_norm": 5.860688209533691,
      "learning_rate": 8.41764841764842e-06,
      "loss": 0.2283,
      "step": 1167
    },
    {
      "epoch": 1.2141372141372142,
      "grad_norm": 5.28604793548584,
      "learning_rate": 8.415338415338416e-06,
      "loss": 0.0811,
      "step": 1168
    },
    {
      "epoch": 1.2151767151767152,
      "grad_norm": 3.432724952697754,
      "learning_rate": 8.413028413028414e-06,
      "loss": 0.1946,
      "step": 1169
    },
    {
      "epoch": 1.2162162162162162,
      "grad_norm": 11.879193305969238,
      "learning_rate": 8.410718410718412e-06,
      "loss": 1.0293,
      "step": 1170
    },
    {
      "epoch": 1.2172557172557172,
      "grad_norm": 4.63136625289917,
      "learning_rate": 8.408408408408409e-06,
      "loss": 0.1425,
      "step": 1171
    },
    {
      "epoch": 1.2182952182952183,
      "grad_norm": 3.035247325897217,
      "learning_rate": 8.406098406098407e-06,
      "loss": 0.0758,
      "step": 1172
    },
    {
      "epoch": 1.2193347193347193,
      "grad_norm": 0.7272508144378662,
      "learning_rate": 8.403788403788405e-06,
      "loss": 0.0207,
      "step": 1173
    },
    {
      "epoch": 1.2203742203742203,
      "grad_norm": 7.765983581542969,
      "learning_rate": 8.401478401478402e-06,
      "loss": 0.4684,
      "step": 1174
    },
    {
      "epoch": 1.2214137214137215,
      "grad_norm": 1.4539945125579834,
      "learning_rate": 8.3991683991684e-06,
      "loss": 0.0477,
      "step": 1175
    },
    {
      "epoch": 1.2224532224532225,
      "grad_norm": 7.17827033996582,
      "learning_rate": 8.396858396858398e-06,
      "loss": 0.5285,
      "step": 1176
    },
    {
      "epoch": 1.2234927234927235,
      "grad_norm": 2.0810439586639404,
      "learning_rate": 8.394548394548394e-06,
      "loss": 0.1235,
      "step": 1177
    },
    {
      "epoch": 1.2245322245322245,
      "grad_norm": 1.930201530456543,
      "learning_rate": 8.392238392238393e-06,
      "loss": 0.0675,
      "step": 1178
    },
    {
      "epoch": 1.2255717255717256,
      "grad_norm": 14.370950698852539,
      "learning_rate": 8.38992838992839e-06,
      "loss": 1.0809,
      "step": 1179
    },
    {
      "epoch": 1.2266112266112266,
      "grad_norm": 7.8660736083984375,
      "learning_rate": 8.387618387618387e-06,
      "loss": 0.5185,
      "step": 1180
    },
    {
      "epoch": 1.2276507276507276,
      "grad_norm": 1.655981183052063,
      "learning_rate": 8.385308385308385e-06,
      "loss": 0.0581,
      "step": 1181
    },
    {
      "epoch": 1.2286902286902288,
      "grad_norm": 3.708282947540283,
      "learning_rate": 8.382998382998384e-06,
      "loss": 0.1657,
      "step": 1182
    },
    {
      "epoch": 1.2297297297297298,
      "grad_norm": 2.6310458183288574,
      "learning_rate": 8.38068838068838e-06,
      "loss": 0.0677,
      "step": 1183
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 2.877119779586792,
      "learning_rate": 8.378378378378378e-06,
      "loss": 0.0854,
      "step": 1184
    },
    {
      "epoch": 1.2318087318087318,
      "grad_norm": 3.463524341583252,
      "learning_rate": 8.376068376068377e-06,
      "loss": 0.1511,
      "step": 1185
    },
    {
      "epoch": 1.2328482328482329,
      "grad_norm": 8.751903533935547,
      "learning_rate": 8.373758373758375e-06,
      "loss": 0.4159,
      "step": 1186
    },
    {
      "epoch": 1.2338877338877339,
      "grad_norm": 0.5461509823799133,
      "learning_rate": 8.371448371448371e-06,
      "loss": 0.0178,
      "step": 1187
    },
    {
      "epoch": 1.2349272349272349,
      "grad_norm": 0.9689798355102539,
      "learning_rate": 8.36913836913837e-06,
      "loss": 0.0244,
      "step": 1188
    },
    {
      "epoch": 1.235966735966736,
      "grad_norm": 8.795180320739746,
      "learning_rate": 8.366828366828368e-06,
      "loss": 0.5102,
      "step": 1189
    },
    {
      "epoch": 1.237006237006237,
      "grad_norm": 1.2706940174102783,
      "learning_rate": 8.364518364518364e-06,
      "loss": 0.0361,
      "step": 1190
    },
    {
      "epoch": 1.2380457380457381,
      "grad_norm": 0.7795925140380859,
      "learning_rate": 8.362208362208362e-06,
      "loss": 0.0183,
      "step": 1191
    },
    {
      "epoch": 1.2390852390852392,
      "grad_norm": 9.270479202270508,
      "learning_rate": 8.35989835989836e-06,
      "loss": 0.8109,
      "step": 1192
    },
    {
      "epoch": 1.2401247401247402,
      "grad_norm": 6.105793476104736,
      "learning_rate": 8.357588357588359e-06,
      "loss": 0.243,
      "step": 1193
    },
    {
      "epoch": 1.2411642411642412,
      "grad_norm": 0.34443458914756775,
      "learning_rate": 8.355278355278355e-06,
      "loss": 0.0044,
      "step": 1194
    },
    {
      "epoch": 1.2422037422037422,
      "grad_norm": 1.8256933689117432,
      "learning_rate": 8.352968352968353e-06,
      "loss": 0.0363,
      "step": 1195
    },
    {
      "epoch": 1.2432432432432432,
      "grad_norm": 8.279906272888184,
      "learning_rate": 8.350658350658352e-06,
      "loss": 0.5154,
      "step": 1196
    },
    {
      "epoch": 1.2442827442827442,
      "grad_norm": 1.0331989526748657,
      "learning_rate": 8.348348348348348e-06,
      "loss": 0.028,
      "step": 1197
    },
    {
      "epoch": 1.2453222453222454,
      "grad_norm": 4.658940315246582,
      "learning_rate": 8.346038346038346e-06,
      "loss": 0.2403,
      "step": 1198
    },
    {
      "epoch": 1.2463617463617465,
      "grad_norm": 5.019296169281006,
      "learning_rate": 8.343728343728344e-06,
      "loss": 0.0645,
      "step": 1199
    },
    {
      "epoch": 1.2474012474012475,
      "grad_norm": 0.07969323545694351,
      "learning_rate": 8.341418341418343e-06,
      "loss": 0.0015,
      "step": 1200
    },
    {
      "epoch": 1.2484407484407485,
      "grad_norm": 0.04548674076795578,
      "learning_rate": 8.339108339108339e-06,
      "loss": 0.0005,
      "step": 1201
    },
    {
      "epoch": 1.2494802494802495,
      "grad_norm": 10.047810554504395,
      "learning_rate": 8.336798336798337e-06,
      "loss": 0.6915,
      "step": 1202
    },
    {
      "epoch": 1.2505197505197505,
      "grad_norm": 0.18876366317272186,
      "learning_rate": 8.334488334488335e-06,
      "loss": 0.005,
      "step": 1203
    },
    {
      "epoch": 1.2515592515592515,
      "grad_norm": 13.021455764770508,
      "learning_rate": 8.332178332178334e-06,
      "loss": 0.9729,
      "step": 1204
    },
    {
      "epoch": 1.2525987525987525,
      "grad_norm": 1.3052574396133423,
      "learning_rate": 8.32986832986833e-06,
      "loss": 0.0346,
      "step": 1205
    },
    {
      "epoch": 1.2536382536382535,
      "grad_norm": 9.151220321655273,
      "learning_rate": 8.327558327558328e-06,
      "loss": 0.5962,
      "step": 1206
    },
    {
      "epoch": 1.2546777546777546,
      "grad_norm": 7.05905294418335,
      "learning_rate": 8.325248325248327e-06,
      "loss": 0.2319,
      "step": 1207
    },
    {
      "epoch": 1.2557172557172558,
      "grad_norm": 6.653399467468262,
      "learning_rate": 8.322938322938323e-06,
      "loss": 0.141,
      "step": 1208
    },
    {
      "epoch": 1.2567567567567568,
      "grad_norm": 0.6559470891952515,
      "learning_rate": 8.320628320628321e-06,
      "loss": 0.0125,
      "step": 1209
    },
    {
      "epoch": 1.2577962577962578,
      "grad_norm": 10.03073501586914,
      "learning_rate": 8.31831831831832e-06,
      "loss": 0.6753,
      "step": 1210
    },
    {
      "epoch": 1.2588357588357588,
      "grad_norm": 1.6613415479660034,
      "learning_rate": 8.316008316008318e-06,
      "loss": 0.0338,
      "step": 1211
    },
    {
      "epoch": 1.2598752598752598,
      "grad_norm": 6.826779842376709,
      "learning_rate": 8.313698313698314e-06,
      "loss": 0.3042,
      "step": 1212
    },
    {
      "epoch": 1.2609147609147608,
      "grad_norm": 2.6513710021972656,
      "learning_rate": 8.311388311388312e-06,
      "loss": 0.1294,
      "step": 1213
    },
    {
      "epoch": 1.261954261954262,
      "grad_norm": 5.151392936706543,
      "learning_rate": 8.30907830907831e-06,
      "loss": 0.0831,
      "step": 1214
    },
    {
      "epoch": 1.262993762993763,
      "grad_norm": 6.245535373687744,
      "learning_rate": 8.306768306768307e-06,
      "loss": 0.1705,
      "step": 1215
    },
    {
      "epoch": 1.264033264033264,
      "grad_norm": 0.5823240876197815,
      "learning_rate": 8.304458304458305e-06,
      "loss": 0.0125,
      "step": 1216
    },
    {
      "epoch": 1.2650727650727651,
      "grad_norm": 9.183516502380371,
      "learning_rate": 8.302148302148303e-06,
      "loss": 0.4556,
      "step": 1217
    },
    {
      "epoch": 1.2661122661122661,
      "grad_norm": 6.539437770843506,
      "learning_rate": 8.299838299838302e-06,
      "loss": 0.2557,
      "step": 1218
    },
    {
      "epoch": 1.2671517671517671,
      "grad_norm": 8.801095962524414,
      "learning_rate": 8.297528297528298e-06,
      "loss": 0.604,
      "step": 1219
    },
    {
      "epoch": 1.2681912681912682,
      "grad_norm": 0.5784547328948975,
      "learning_rate": 8.295218295218296e-06,
      "loss": 0.0208,
      "step": 1220
    },
    {
      "epoch": 1.2692307692307692,
      "grad_norm": 7.860816955566406,
      "learning_rate": 8.292908292908294e-06,
      "loss": 0.329,
      "step": 1221
    },
    {
      "epoch": 1.2702702702702702,
      "grad_norm": 0.34168723225593567,
      "learning_rate": 8.290598290598293e-06,
      "loss": 0.008,
      "step": 1222
    },
    {
      "epoch": 1.2713097713097712,
      "grad_norm": 3.0554420948028564,
      "learning_rate": 8.288288288288289e-06,
      "loss": 0.0614,
      "step": 1223
    },
    {
      "epoch": 1.2723492723492724,
      "grad_norm": 13.036967277526855,
      "learning_rate": 8.285978285978287e-06,
      "loss": 0.7422,
      "step": 1224
    },
    {
      "epoch": 1.2733887733887734,
      "grad_norm": 5.609252452850342,
      "learning_rate": 8.283668283668286e-06,
      "loss": 0.2516,
      "step": 1225
    },
    {
      "epoch": 1.2744282744282744,
      "grad_norm": 10.398143768310547,
      "learning_rate": 8.281358281358282e-06,
      "loss": 0.7848,
      "step": 1226
    },
    {
      "epoch": 1.2754677754677755,
      "grad_norm": 9.209569931030273,
      "learning_rate": 8.27904827904828e-06,
      "loss": 0.1943,
      "step": 1227
    },
    {
      "epoch": 1.2765072765072765,
      "grad_norm": 7.4310197830200195,
      "learning_rate": 8.276738276738278e-06,
      "loss": 0.2437,
      "step": 1228
    },
    {
      "epoch": 1.2775467775467775,
      "grad_norm": 11.599946022033691,
      "learning_rate": 8.274428274428275e-06,
      "loss": 0.7414,
      "step": 1229
    },
    {
      "epoch": 1.2785862785862787,
      "grad_norm": 13.055266380310059,
      "learning_rate": 8.272118272118273e-06,
      "loss": 0.6268,
      "step": 1230
    },
    {
      "epoch": 1.2796257796257797,
      "grad_norm": 10.34594440460205,
      "learning_rate": 8.269808269808271e-06,
      "loss": 0.3585,
      "step": 1231
    },
    {
      "epoch": 1.2806652806652807,
      "grad_norm": 4.395898818969727,
      "learning_rate": 8.267498267498268e-06,
      "loss": 0.0964,
      "step": 1232
    },
    {
      "epoch": 1.2817047817047817,
      "grad_norm": 3.7805607318878174,
      "learning_rate": 8.265188265188266e-06,
      "loss": 0.1501,
      "step": 1233
    },
    {
      "epoch": 1.2827442827442828,
      "grad_norm": 8.583539009094238,
      "learning_rate": 8.262878262878264e-06,
      "loss": 0.6436,
      "step": 1234
    },
    {
      "epoch": 1.2837837837837838,
      "grad_norm": 2.4144604206085205,
      "learning_rate": 8.26056826056826e-06,
      "loss": 0.0688,
      "step": 1235
    },
    {
      "epoch": 1.2848232848232848,
      "grad_norm": 4.065729141235352,
      "learning_rate": 8.258258258258259e-06,
      "loss": 0.073,
      "step": 1236
    },
    {
      "epoch": 1.2858627858627858,
      "grad_norm": 4.701965808868408,
      "learning_rate": 8.255948255948257e-06,
      "loss": 0.1044,
      "step": 1237
    },
    {
      "epoch": 1.2869022869022868,
      "grad_norm": 3.801809787750244,
      "learning_rate": 8.253638253638254e-06,
      "loss": 0.1211,
      "step": 1238
    },
    {
      "epoch": 1.2879417879417878,
      "grad_norm": 9.999377250671387,
      "learning_rate": 8.251328251328252e-06,
      "loss": 0.5039,
      "step": 1239
    },
    {
      "epoch": 1.288981288981289,
      "grad_norm": 10.668828964233398,
      "learning_rate": 8.24901824901825e-06,
      "loss": 0.5704,
      "step": 1240
    },
    {
      "epoch": 1.29002079002079,
      "grad_norm": 7.582046985626221,
      "learning_rate": 8.246708246708246e-06,
      "loss": 0.3647,
      "step": 1241
    },
    {
      "epoch": 1.291060291060291,
      "grad_norm": 6.662896633148193,
      "learning_rate": 8.244398244398245e-06,
      "loss": 0.1706,
      "step": 1242
    },
    {
      "epoch": 1.292099792099792,
      "grad_norm": 14.700654029846191,
      "learning_rate": 8.242088242088243e-06,
      "loss": 0.9503,
      "step": 1243
    },
    {
      "epoch": 1.293139293139293,
      "grad_norm": 0.9491019248962402,
      "learning_rate": 8.23977823977824e-06,
      "loss": 0.0185,
      "step": 1244
    },
    {
      "epoch": 1.2941787941787941,
      "grad_norm": 7.621535778045654,
      "learning_rate": 8.237468237468237e-06,
      "loss": 0.2511,
      "step": 1245
    },
    {
      "epoch": 1.2952182952182953,
      "grad_norm": 4.274621486663818,
      "learning_rate": 8.235158235158236e-06,
      "loss": 0.061,
      "step": 1246
    },
    {
      "epoch": 1.2962577962577964,
      "grad_norm": 13.053854942321777,
      "learning_rate": 8.232848232848234e-06,
      "loss": 0.5283,
      "step": 1247
    },
    {
      "epoch": 1.2972972972972974,
      "grad_norm": 13.249424934387207,
      "learning_rate": 8.23053823053823e-06,
      "loss": 0.5875,
      "step": 1248
    },
    {
      "epoch": 1.2983367983367984,
      "grad_norm": 9.532174110412598,
      "learning_rate": 8.228228228228229e-06,
      "loss": 0.231,
      "step": 1249
    },
    {
      "epoch": 1.2993762993762994,
      "grad_norm": 3.8693883419036865,
      "learning_rate": 8.225918225918227e-06,
      "loss": 0.1521,
      "step": 1250
    },
    {
      "epoch": 1.3004158004158004,
      "grad_norm": 9.420363426208496,
      "learning_rate": 8.223608223608223e-06,
      "loss": 0.405,
      "step": 1251
    },
    {
      "epoch": 1.3014553014553014,
      "grad_norm": 6.109569072723389,
      "learning_rate": 8.221298221298221e-06,
      "loss": 0.2135,
      "step": 1252
    },
    {
      "epoch": 1.3024948024948024,
      "grad_norm": 13.657364845275879,
      "learning_rate": 8.21898821898822e-06,
      "loss": 0.9535,
      "step": 1253
    },
    {
      "epoch": 1.3035343035343034,
      "grad_norm": 9.316267013549805,
      "learning_rate": 8.216678216678218e-06,
      "loss": 0.368,
      "step": 1254
    },
    {
      "epoch": 1.3045738045738045,
      "grad_norm": 5.461922645568848,
      "learning_rate": 8.214368214368214e-06,
      "loss": 0.1893,
      "step": 1255
    },
    {
      "epoch": 1.3056133056133057,
      "grad_norm": 0.023014483973383904,
      "learning_rate": 8.212058212058212e-06,
      "loss": 0.0005,
      "step": 1256
    },
    {
      "epoch": 1.3066528066528067,
      "grad_norm": 0.6040786504745483,
      "learning_rate": 8.20974820974821e-06,
      "loss": 0.0146,
      "step": 1257
    },
    {
      "epoch": 1.3076923076923077,
      "grad_norm": 2.6426706314086914,
      "learning_rate": 8.207438207438209e-06,
      "loss": 0.0522,
      "step": 1258
    },
    {
      "epoch": 1.3087318087318087,
      "grad_norm": 0.6564608812332153,
      "learning_rate": 8.205128205128205e-06,
      "loss": 0.0132,
      "step": 1259
    },
    {
      "epoch": 1.3097713097713097,
      "grad_norm": 9.825348854064941,
      "learning_rate": 8.202818202818204e-06,
      "loss": 0.3831,
      "step": 1260
    },
    {
      "epoch": 1.3108108108108107,
      "grad_norm": 9.027199745178223,
      "learning_rate": 8.200508200508202e-06,
      "loss": 0.1866,
      "step": 1261
    },
    {
      "epoch": 1.311850311850312,
      "grad_norm": 12.532243728637695,
      "learning_rate": 8.198198198198198e-06,
      "loss": 1.3439,
      "step": 1262
    },
    {
      "epoch": 1.312889812889813,
      "grad_norm": 7.945847511291504,
      "learning_rate": 8.195888195888196e-06,
      "loss": 0.0877,
      "step": 1263
    },
    {
      "epoch": 1.313929313929314,
      "grad_norm": 9.476608276367188,
      "learning_rate": 8.193578193578195e-06,
      "loss": 0.4055,
      "step": 1264
    },
    {
      "epoch": 1.314968814968815,
      "grad_norm": 8.13533878326416,
      "learning_rate": 8.191268191268193e-06,
      "loss": 0.1765,
      "step": 1265
    },
    {
      "epoch": 1.316008316008316,
      "grad_norm": 1.300158977508545,
      "learning_rate": 8.18895818895819e-06,
      "loss": 0.0216,
      "step": 1266
    },
    {
      "epoch": 1.317047817047817,
      "grad_norm": 2.6788394451141357,
      "learning_rate": 8.186648186648188e-06,
      "loss": 0.09,
      "step": 1267
    },
    {
      "epoch": 1.318087318087318,
      "grad_norm": 4.26913595199585,
      "learning_rate": 8.184338184338186e-06,
      "loss": 0.2048,
      "step": 1268
    },
    {
      "epoch": 1.319126819126819,
      "grad_norm": 15.426655769348145,
      "learning_rate": 8.182028182028182e-06,
      "loss": 1.023,
      "step": 1269
    },
    {
      "epoch": 1.32016632016632,
      "grad_norm": 8.184285163879395,
      "learning_rate": 8.17971817971818e-06,
      "loss": 0.2651,
      "step": 1270
    },
    {
      "epoch": 1.321205821205821,
      "grad_norm": 0.1392313688993454,
      "learning_rate": 8.177408177408179e-06,
      "loss": 0.0016,
      "step": 1271
    },
    {
      "epoch": 1.3222453222453223,
      "grad_norm": 0.1190238893032074,
      "learning_rate": 8.175098175098177e-06,
      "loss": 0.0021,
      "step": 1272
    },
    {
      "epoch": 1.3232848232848233,
      "grad_norm": 17.05913734436035,
      "learning_rate": 8.172788172788173e-06,
      "loss": 1.2166,
      "step": 1273
    },
    {
      "epoch": 1.3243243243243243,
      "grad_norm": 0.30950024724006653,
      "learning_rate": 8.170478170478171e-06,
      "loss": 0.0065,
      "step": 1274
    },
    {
      "epoch": 1.3253638253638254,
      "grad_norm": 1.1400848627090454,
      "learning_rate": 8.16816816816817e-06,
      "loss": 0.0196,
      "step": 1275
    },
    {
      "epoch": 1.3264033264033264,
      "grad_norm": 8.545634269714355,
      "learning_rate": 8.165858165858168e-06,
      "loss": 0.3026,
      "step": 1276
    },
    {
      "epoch": 1.3274428274428274,
      "grad_norm": 0.4845014810562134,
      "learning_rate": 8.163548163548164e-06,
      "loss": 0.006,
      "step": 1277
    },
    {
      "epoch": 1.3284823284823286,
      "grad_norm": 16.63973045349121,
      "learning_rate": 8.161238161238163e-06,
      "loss": 1.0234,
      "step": 1278
    },
    {
      "epoch": 1.3295218295218296,
      "grad_norm": 15.797550201416016,
      "learning_rate": 8.15892815892816e-06,
      "loss": 1.2203,
      "step": 1279
    },
    {
      "epoch": 1.3305613305613306,
      "grad_norm": 1.719501256942749,
      "learning_rate": 8.156618156618157e-06,
      "loss": 0.0417,
      "step": 1280
    },
    {
      "epoch": 1.3316008316008316,
      "grad_norm": 16.285903930664062,
      "learning_rate": 8.154308154308155e-06,
      "loss": 0.2835,
      "step": 1281
    },
    {
      "epoch": 1.3326403326403327,
      "grad_norm": 0.5948153138160706,
      "learning_rate": 8.151998151998154e-06,
      "loss": 0.0117,
      "step": 1282
    },
    {
      "epoch": 1.3336798336798337,
      "grad_norm": 13.316926956176758,
      "learning_rate": 8.14968814968815e-06,
      "loss": 0.8549,
      "step": 1283
    },
    {
      "epoch": 1.3347193347193347,
      "grad_norm": 6.364133834838867,
      "learning_rate": 8.147378147378148e-06,
      "loss": 0.3454,
      "step": 1284
    },
    {
      "epoch": 1.3357588357588357,
      "grad_norm": 8.6270170211792,
      "learning_rate": 8.145068145068146e-06,
      "loss": 0.4939,
      "step": 1285
    },
    {
      "epoch": 1.3367983367983367,
      "grad_norm": 0.4271339774131775,
      "learning_rate": 8.142758142758143e-06,
      "loss": 0.0111,
      "step": 1286
    },
    {
      "epoch": 1.3378378378378377,
      "grad_norm": 10.21527099609375,
      "learning_rate": 8.140448140448141e-06,
      "loss": 0.5921,
      "step": 1287
    },
    {
      "epoch": 1.338877338877339,
      "grad_norm": 10.404336929321289,
      "learning_rate": 8.13813813813814e-06,
      "loss": 0.6905,
      "step": 1288
    },
    {
      "epoch": 1.33991683991684,
      "grad_norm": 1.3672646284103394,
      "learning_rate": 8.135828135828136e-06,
      "loss": 0.0269,
      "step": 1289
    },
    {
      "epoch": 1.340956340956341,
      "grad_norm": 1.9783116579055786,
      "learning_rate": 8.133518133518134e-06,
      "loss": 0.0359,
      "step": 1290
    },
    {
      "epoch": 1.341995841995842,
      "grad_norm": 4.706028938293457,
      "learning_rate": 8.131208131208132e-06,
      "loss": 0.1376,
      "step": 1291
    },
    {
      "epoch": 1.343035343035343,
      "grad_norm": 6.213561058044434,
      "learning_rate": 8.128898128898129e-06,
      "loss": 0.1551,
      "step": 1292
    },
    {
      "epoch": 1.344074844074844,
      "grad_norm": 17.572166442871094,
      "learning_rate": 8.126588126588127e-06,
      "loss": 0.1122,
      "step": 1293
    },
    {
      "epoch": 1.3451143451143452,
      "grad_norm": 9.871464729309082,
      "learning_rate": 8.124278124278125e-06,
      "loss": 0.4517,
      "step": 1294
    },
    {
      "epoch": 1.3461538461538463,
      "grad_norm": 3.3636231422424316,
      "learning_rate": 8.121968121968122e-06,
      "loss": 0.1099,
      "step": 1295
    },
    {
      "epoch": 1.3471933471933473,
      "grad_norm": 6.92306661605835,
      "learning_rate": 8.11965811965812e-06,
      "loss": 0.2374,
      "step": 1296
    },
    {
      "epoch": 1.3482328482328483,
      "grad_norm": 12.991156578063965,
      "learning_rate": 8.117348117348118e-06,
      "loss": 1.7004,
      "step": 1297
    },
    {
      "epoch": 1.3492723492723493,
      "grad_norm": 2.913350820541382,
      "learning_rate": 8.115038115038114e-06,
      "loss": 0.0435,
      "step": 1298
    },
    {
      "epoch": 1.3503118503118503,
      "grad_norm": 0.5244566798210144,
      "learning_rate": 8.112728112728113e-06,
      "loss": 0.0153,
      "step": 1299
    },
    {
      "epoch": 1.3513513513513513,
      "grad_norm": 11.311691284179688,
      "learning_rate": 8.110418110418111e-06,
      "loss": 0.1636,
      "step": 1300
    },
    {
      "epoch": 1.3523908523908523,
      "grad_norm": 11.263629913330078,
      "learning_rate": 8.108108108108109e-06,
      "loss": 0.4313,
      "step": 1301
    },
    {
      "epoch": 1.3534303534303533,
      "grad_norm": 2.7492759227752686,
      "learning_rate": 8.105798105798106e-06,
      "loss": 0.0683,
      "step": 1302
    },
    {
      "epoch": 1.3544698544698544,
      "grad_norm": 2.5676145553588867,
      "learning_rate": 8.103488103488104e-06,
      "loss": 0.097,
      "step": 1303
    },
    {
      "epoch": 1.3555093555093556,
      "grad_norm": 3.0315756797790527,
      "learning_rate": 8.101178101178102e-06,
      "loss": 0.2299,
      "step": 1304
    },
    {
      "epoch": 1.3565488565488566,
      "grad_norm": 10.284161567687988,
      "learning_rate": 8.098868098868098e-06,
      "loss": 1.137,
      "step": 1305
    },
    {
      "epoch": 1.3575883575883576,
      "grad_norm": 0.3827531039714813,
      "learning_rate": 8.096558096558097e-06,
      "loss": 0.0108,
      "step": 1306
    },
    {
      "epoch": 1.3586278586278586,
      "grad_norm": 4.843155384063721,
      "learning_rate": 8.094248094248095e-06,
      "loss": 0.06,
      "step": 1307
    },
    {
      "epoch": 1.3596673596673596,
      "grad_norm": 13.512953758239746,
      "learning_rate": 8.091938091938093e-06,
      "loss": 1.04,
      "step": 1308
    },
    {
      "epoch": 1.3607068607068606,
      "grad_norm": 5.828279972076416,
      "learning_rate": 8.08962808962809e-06,
      "loss": 0.089,
      "step": 1309
    },
    {
      "epoch": 1.3617463617463619,
      "grad_norm": 2.431767702102661,
      "learning_rate": 8.087318087318088e-06,
      "loss": 0.0354,
      "step": 1310
    },
    {
      "epoch": 1.362785862785863,
      "grad_norm": 1.5293114185333252,
      "learning_rate": 8.085008085008086e-06,
      "loss": 0.0218,
      "step": 1311
    },
    {
      "epoch": 1.363825363825364,
      "grad_norm": 11.862726211547852,
      "learning_rate": 8.082698082698084e-06,
      "loss": 0.4311,
      "step": 1312
    },
    {
      "epoch": 1.364864864864865,
      "grad_norm": 14.451634407043457,
      "learning_rate": 8.08038808038808e-06,
      "loss": 2.1098,
      "step": 1313
    },
    {
      "epoch": 1.365904365904366,
      "grad_norm": 11.710643768310547,
      "learning_rate": 8.078078078078079e-06,
      "loss": 0.2682,
      "step": 1314
    },
    {
      "epoch": 1.366943866943867,
      "grad_norm": 1.511605143547058,
      "learning_rate": 8.075768075768077e-06,
      "loss": 0.0272,
      "step": 1315
    },
    {
      "epoch": 1.367983367983368,
      "grad_norm": 11.279335021972656,
      "learning_rate": 8.073458073458073e-06,
      "loss": 0.7856,
      "step": 1316
    },
    {
      "epoch": 1.369022869022869,
      "grad_norm": 2.178616762161255,
      "learning_rate": 8.071148071148072e-06,
      "loss": 0.0562,
      "step": 1317
    },
    {
      "epoch": 1.37006237006237,
      "grad_norm": 1.0928305387496948,
      "learning_rate": 8.06883806883807e-06,
      "loss": 0.0292,
      "step": 1318
    },
    {
      "epoch": 1.371101871101871,
      "grad_norm": 0.34847453236579895,
      "learning_rate": 8.066528066528068e-06,
      "loss": 0.0067,
      "step": 1319
    },
    {
      "epoch": 1.3721413721413722,
      "grad_norm": 0.4622173011302948,
      "learning_rate": 8.064218064218065e-06,
      "loss": 0.0092,
      "step": 1320
    },
    {
      "epoch": 1.3731808731808732,
      "grad_norm": 0.11356835067272186,
      "learning_rate": 8.061908061908063e-06,
      "loss": 0.0021,
      "step": 1321
    },
    {
      "epoch": 1.3742203742203742,
      "grad_norm": 0.6464567184448242,
      "learning_rate": 8.059598059598061e-06,
      "loss": 0.0167,
      "step": 1322
    },
    {
      "epoch": 1.3752598752598753,
      "grad_norm": 9.651885032653809,
      "learning_rate": 8.057288057288057e-06,
      "loss": 0.2491,
      "step": 1323
    },
    {
      "epoch": 1.3762993762993763,
      "grad_norm": 3.052480936050415,
      "learning_rate": 8.054978054978056e-06,
      "loss": 0.1249,
      "step": 1324
    },
    {
      "epoch": 1.3773388773388773,
      "grad_norm": 2.6563804149627686,
      "learning_rate": 8.052668052668054e-06,
      "loss": 0.0743,
      "step": 1325
    },
    {
      "epoch": 1.3783783783783785,
      "grad_norm": 2.6560709476470947,
      "learning_rate": 8.050358050358052e-06,
      "loss": 0.0542,
      "step": 1326
    },
    {
      "epoch": 1.3794178794178795,
      "grad_norm": 7.124080657958984,
      "learning_rate": 8.048048048048048e-06,
      "loss": 0.2983,
      "step": 1327
    },
    {
      "epoch": 1.3804573804573805,
      "grad_norm": 5.484076499938965,
      "learning_rate": 8.045738045738047e-06,
      "loss": 0.0964,
      "step": 1328
    },
    {
      "epoch": 1.3814968814968815,
      "grad_norm": 4.23430061340332,
      "learning_rate": 8.043428043428045e-06,
      "loss": 0.045,
      "step": 1329
    },
    {
      "epoch": 1.3825363825363826,
      "grad_norm": 0.15918128192424774,
      "learning_rate": 8.041118041118043e-06,
      "loss": 0.0051,
      "step": 1330
    },
    {
      "epoch": 1.3835758835758836,
      "grad_norm": 9.001898765563965,
      "learning_rate": 8.03880803880804e-06,
      "loss": 0.3259,
      "step": 1331
    },
    {
      "epoch": 1.3846153846153846,
      "grad_norm": 7.946660041809082,
      "learning_rate": 8.036498036498038e-06,
      "loss": 0.3321,
      "step": 1332
    },
    {
      "epoch": 1.3856548856548856,
      "grad_norm": 9.039274215698242,
      "learning_rate": 8.034188034188036e-06,
      "loss": 0.1995,
      "step": 1333
    },
    {
      "epoch": 1.3866943866943866,
      "grad_norm": 9.994538307189941,
      "learning_rate": 8.031878031878032e-06,
      "loss": 0.7766,
      "step": 1334
    },
    {
      "epoch": 1.3877338877338876,
      "grad_norm": 0.012579978443682194,
      "learning_rate": 8.02956802956803e-06,
      "loss": 0.0001,
      "step": 1335
    },
    {
      "epoch": 1.3887733887733889,
      "grad_norm": 10.88592529296875,
      "learning_rate": 8.027258027258029e-06,
      "loss": 1.4343,
      "step": 1336
    },
    {
      "epoch": 1.3898128898128899,
      "grad_norm": 2.3097336292266846,
      "learning_rate": 8.024948024948025e-06,
      "loss": 0.0364,
      "step": 1337
    },
    {
      "epoch": 1.3908523908523909,
      "grad_norm": 7.158543586730957,
      "learning_rate": 8.022638022638023e-06,
      "loss": 0.1027,
      "step": 1338
    },
    {
      "epoch": 1.3918918918918919,
      "grad_norm": 10.87025260925293,
      "learning_rate": 8.020328020328022e-06,
      "loss": 0.3478,
      "step": 1339
    },
    {
      "epoch": 1.392931392931393,
      "grad_norm": 0.6269477605819702,
      "learning_rate": 8.018018018018018e-06,
      "loss": 0.0067,
      "step": 1340
    },
    {
      "epoch": 1.393970893970894,
      "grad_norm": 0.879704475402832,
      "learning_rate": 8.015708015708016e-06,
      "loss": 0.0216,
      "step": 1341
    },
    {
      "epoch": 1.3950103950103951,
      "grad_norm": 0.16545157134532928,
      "learning_rate": 8.013398013398015e-06,
      "loss": 0.0041,
      "step": 1342
    },
    {
      "epoch": 1.3960498960498962,
      "grad_norm": 2.778501272201538,
      "learning_rate": 8.011088011088011e-06,
      "loss": 0.0508,
      "step": 1343
    },
    {
      "epoch": 1.3970893970893972,
      "grad_norm": 3.787515878677368,
      "learning_rate": 8.00877800877801e-06,
      "loss": 0.0563,
      "step": 1344
    },
    {
      "epoch": 1.3981288981288982,
      "grad_norm": 0.3681766390800476,
      "learning_rate": 8.006468006468007e-06,
      "loss": 0.0086,
      "step": 1345
    },
    {
      "epoch": 1.3991683991683992,
      "grad_norm": 2.6923065185546875,
      "learning_rate": 8.004158004158004e-06,
      "loss": 0.0508,
      "step": 1346
    },
    {
      "epoch": 1.4002079002079002,
      "grad_norm": 5.058040142059326,
      "learning_rate": 8.001848001848002e-06,
      "loss": 0.0971,
      "step": 1347
    },
    {
      "epoch": 1.4012474012474012,
      "grad_norm": 5.924242973327637,
      "learning_rate": 7.999537999538e-06,
      "loss": 0.4377,
      "step": 1348
    },
    {
      "epoch": 1.4022869022869022,
      "grad_norm": 0.38952162861824036,
      "learning_rate": 7.997227997227997e-06,
      "loss": 0.0083,
      "step": 1349
    },
    {
      "epoch": 1.4033264033264032,
      "grad_norm": 11.313742637634277,
      "learning_rate": 7.994917994917995e-06,
      "loss": 0.9199,
      "step": 1350
    },
    {
      "epoch": 1.4043659043659042,
      "grad_norm": 11.081106185913086,
      "learning_rate": 7.992607992607993e-06,
      "loss": 0.5341,
      "step": 1351
    },
    {
      "epoch": 1.4054054054054055,
      "grad_norm": 18.970726013183594,
      "learning_rate": 7.99029799029799e-06,
      "loss": 1.0361,
      "step": 1352
    },
    {
      "epoch": 1.4064449064449065,
      "grad_norm": 8.993247985839844,
      "learning_rate": 7.987987987987988e-06,
      "loss": 0.7547,
      "step": 1353
    },
    {
      "epoch": 1.4074844074844075,
      "grad_norm": 0.08616983145475388,
      "learning_rate": 7.985677985677986e-06,
      "loss": 0.0015,
      "step": 1354
    },
    {
      "epoch": 1.4085239085239085,
      "grad_norm": 9.323349952697754,
      "learning_rate": 7.983367983367984e-06,
      "loss": 0.2254,
      "step": 1355
    },
    {
      "epoch": 1.4095634095634095,
      "grad_norm": 1.1458944082260132,
      "learning_rate": 7.98105798105798e-06,
      "loss": 0.051,
      "step": 1356
    },
    {
      "epoch": 1.4106029106029105,
      "grad_norm": 22.63304901123047,
      "learning_rate": 7.978747978747979e-06,
      "loss": 0.6127,
      "step": 1357
    },
    {
      "epoch": 1.4116424116424118,
      "grad_norm": 2.062016725540161,
      "learning_rate": 7.976437976437977e-06,
      "loss": 0.0247,
      "step": 1358
    },
    {
      "epoch": 1.4126819126819128,
      "grad_norm": 1.6762837171554565,
      "learning_rate": 7.974127974127974e-06,
      "loss": 0.0488,
      "step": 1359
    },
    {
      "epoch": 1.4137214137214138,
      "grad_norm": 2.0908496379852295,
      "learning_rate": 7.971817971817972e-06,
      "loss": 0.0375,
      "step": 1360
    },
    {
      "epoch": 1.4147609147609148,
      "grad_norm": 1.9176737070083618,
      "learning_rate": 7.96950796950797e-06,
      "loss": 0.0213,
      "step": 1361
    },
    {
      "epoch": 1.4158004158004158,
      "grad_norm": 8.565489768981934,
      "learning_rate": 7.967197967197968e-06,
      "loss": 0.3279,
      "step": 1362
    },
    {
      "epoch": 1.4168399168399168,
      "grad_norm": 6.474407196044922,
      "learning_rate": 7.964887964887965e-06,
      "loss": 0.1724,
      "step": 1363
    },
    {
      "epoch": 1.4178794178794178,
      "grad_norm": 10.128867149353027,
      "learning_rate": 7.962577962577963e-06,
      "loss": 0.3057,
      "step": 1364
    },
    {
      "epoch": 1.4189189189189189,
      "grad_norm": 1.244747281074524,
      "learning_rate": 7.960267960267961e-06,
      "loss": 0.0178,
      "step": 1365
    },
    {
      "epoch": 1.4199584199584199,
      "grad_norm": 10.41561508178711,
      "learning_rate": 7.95795795795796e-06,
      "loss": 0.3249,
      "step": 1366
    },
    {
      "epoch": 1.4209979209979209,
      "grad_norm": 6.584688186645508,
      "learning_rate": 7.955647955647956e-06,
      "loss": 0.1788,
      "step": 1367
    },
    {
      "epoch": 1.4220374220374221,
      "grad_norm": 0.3154916763305664,
      "learning_rate": 7.953337953337954e-06,
      "loss": 0.0071,
      "step": 1368
    },
    {
      "epoch": 1.4230769230769231,
      "grad_norm": 7.990124225616455,
      "learning_rate": 7.951027951027952e-06,
      "loss": 0.2696,
      "step": 1369
    },
    {
      "epoch": 1.4241164241164241,
      "grad_norm": 0.6271470189094543,
      "learning_rate": 7.948717948717949e-06,
      "loss": 0.0121,
      "step": 1370
    },
    {
      "epoch": 1.4251559251559252,
      "grad_norm": 11.38403606414795,
      "learning_rate": 7.946407946407947e-06,
      "loss": 0.4569,
      "step": 1371
    },
    {
      "epoch": 1.4261954261954262,
      "grad_norm": 0.29966673254966736,
      "learning_rate": 7.944097944097945e-06,
      "loss": 0.0057,
      "step": 1372
    },
    {
      "epoch": 1.4272349272349272,
      "grad_norm": 3.51631498336792,
      "learning_rate": 7.941787941787943e-06,
      "loss": 0.0646,
      "step": 1373
    },
    {
      "epoch": 1.4282744282744284,
      "grad_norm": 0.11474553495645523,
      "learning_rate": 7.93947793947794e-06,
      "loss": 0.0014,
      "step": 1374
    },
    {
      "epoch": 1.4293139293139294,
      "grad_norm": 0.058466557413339615,
      "learning_rate": 7.937167937167938e-06,
      "loss": 0.0007,
      "step": 1375
    },
    {
      "epoch": 1.4303534303534304,
      "grad_norm": 2.909372568130493,
      "learning_rate": 7.934857934857936e-06,
      "loss": 0.0337,
      "step": 1376
    },
    {
      "epoch": 1.4313929313929314,
      "grad_norm": 2.4320931434631348,
      "learning_rate": 7.932547932547933e-06,
      "loss": 0.0398,
      "step": 1377
    },
    {
      "epoch": 1.4324324324324325,
      "grad_norm": 6.761545658111572,
      "learning_rate": 7.93023793023793e-06,
      "loss": 0.2224,
      "step": 1378
    },
    {
      "epoch": 1.4334719334719335,
      "grad_norm": 3.9421579837799072,
      "learning_rate": 7.927927927927929e-06,
      "loss": 0.0674,
      "step": 1379
    },
    {
      "epoch": 1.4345114345114345,
      "grad_norm": 3.7583696842193604,
      "learning_rate": 7.925617925617927e-06,
      "loss": 0.1516,
      "step": 1380
    },
    {
      "epoch": 1.4355509355509355,
      "grad_norm": 4.613794326782227,
      "learning_rate": 7.923307923307924e-06,
      "loss": 0.0488,
      "step": 1381
    },
    {
      "epoch": 1.4365904365904365,
      "grad_norm": 7.776555061340332,
      "learning_rate": 7.920997920997922e-06,
      "loss": 0.4145,
      "step": 1382
    },
    {
      "epoch": 1.4376299376299375,
      "grad_norm": 10.442293167114258,
      "learning_rate": 7.91868791868792e-06,
      "loss": 0.8719,
      "step": 1383
    },
    {
      "epoch": 1.4386694386694387,
      "grad_norm": 0.04214008152484894,
      "learning_rate": 7.916377916377918e-06,
      "loss": 0.0009,
      "step": 1384
    },
    {
      "epoch": 1.4397089397089398,
      "grad_norm": 0.6979100704193115,
      "learning_rate": 7.914067914067915e-06,
      "loss": 0.0195,
      "step": 1385
    },
    {
      "epoch": 1.4407484407484408,
      "grad_norm": 6.32226037979126,
      "learning_rate": 7.911757911757913e-06,
      "loss": 0.216,
      "step": 1386
    },
    {
      "epoch": 1.4417879417879418,
      "grad_norm": 5.914670944213867,
      "learning_rate": 7.909447909447911e-06,
      "loss": 0.1529,
      "step": 1387
    },
    {
      "epoch": 1.4428274428274428,
      "grad_norm": 2.56058669090271,
      "learning_rate": 7.907137907137908e-06,
      "loss": 0.0471,
      "step": 1388
    },
    {
      "epoch": 1.4438669438669438,
      "grad_norm": 12.25080680847168,
      "learning_rate": 7.904827904827906e-06,
      "loss": 0.4424,
      "step": 1389
    },
    {
      "epoch": 1.444906444906445,
      "grad_norm": 5.598354339599609,
      "learning_rate": 7.902517902517904e-06,
      "loss": 0.1845,
      "step": 1390
    },
    {
      "epoch": 1.445945945945946,
      "grad_norm": 0.8475781083106995,
      "learning_rate": 7.9002079002079e-06,
      "loss": 0.0107,
      "step": 1391
    },
    {
      "epoch": 1.446985446985447,
      "grad_norm": 12.017598152160645,
      "learning_rate": 7.897897897897899e-06,
      "loss": 0.9019,
      "step": 1392
    },
    {
      "epoch": 1.448024948024948,
      "grad_norm": 26.395370483398438,
      "learning_rate": 7.895587895587897e-06,
      "loss": 0.6365,
      "step": 1393
    },
    {
      "epoch": 1.449064449064449,
      "grad_norm": 8.331962585449219,
      "learning_rate": 7.893277893277893e-06,
      "loss": 0.2966,
      "step": 1394
    },
    {
      "epoch": 1.45010395010395,
      "grad_norm": 0.055291905999183655,
      "learning_rate": 7.890967890967892e-06,
      "loss": 0.0011,
      "step": 1395
    },
    {
      "epoch": 1.4511434511434511,
      "grad_norm": 15.261618614196777,
      "learning_rate": 7.88865788865789e-06,
      "loss": 0.7521,
      "step": 1396
    },
    {
      "epoch": 1.4521829521829521,
      "grad_norm": 0.3626764416694641,
      "learning_rate": 7.886347886347886e-06,
      "loss": 0.0083,
      "step": 1397
    },
    {
      "epoch": 1.4532224532224531,
      "grad_norm": 8.048782348632812,
      "learning_rate": 7.884037884037884e-06,
      "loss": 0.502,
      "step": 1398
    },
    {
      "epoch": 1.4542619542619541,
      "grad_norm": 0.0779711902141571,
      "learning_rate": 7.881727881727883e-06,
      "loss": 0.002,
      "step": 1399
    },
    {
      "epoch": 1.4553014553014554,
      "grad_norm": 9.601903915405273,
      "learning_rate": 7.879417879417879e-06,
      "loss": 0.1312,
      "step": 1400
    },
    {
      "epoch": 1.4563409563409564,
      "grad_norm": 9.24765682220459,
      "learning_rate": 7.877107877107877e-06,
      "loss": 0.3113,
      "step": 1401
    },
    {
      "epoch": 1.4573804573804574,
      "grad_norm": 13.755073547363281,
      "learning_rate": 7.874797874797875e-06,
      "loss": 0.5496,
      "step": 1402
    },
    {
      "epoch": 1.4584199584199584,
      "grad_norm": 6.341691493988037,
      "learning_rate": 7.872487872487872e-06,
      "loss": 0.0982,
      "step": 1403
    },
    {
      "epoch": 1.4594594594594594,
      "grad_norm": 19.51810646057129,
      "learning_rate": 7.87017787017787e-06,
      "loss": 0.5321,
      "step": 1404
    },
    {
      "epoch": 1.4604989604989604,
      "grad_norm": 1.2743085622787476,
      "learning_rate": 7.867867867867868e-06,
      "loss": 0.0373,
      "step": 1405
    },
    {
      "epoch": 1.4615384615384617,
      "grad_norm": 1.669488549232483,
      "learning_rate": 7.865557865557867e-06,
      "loss": 0.0329,
      "step": 1406
    },
    {
      "epoch": 1.4625779625779627,
      "grad_norm": 11.852375030517578,
      "learning_rate": 7.863247863247863e-06,
      "loss": 0.8486,
      "step": 1407
    },
    {
      "epoch": 1.4636174636174637,
      "grad_norm": 4.6349568367004395,
      "learning_rate": 7.860937860937861e-06,
      "loss": 0.1185,
      "step": 1408
    },
    {
      "epoch": 1.4646569646569647,
      "grad_norm": 3.4211432933807373,
      "learning_rate": 7.85862785862786e-06,
      "loss": 0.1006,
      "step": 1409
    },
    {
      "epoch": 1.4656964656964657,
      "grad_norm": 3.819972038269043,
      "learning_rate": 7.856317856317856e-06,
      "loss": 0.0653,
      "step": 1410
    },
    {
      "epoch": 1.4667359667359667,
      "grad_norm": 2.380999803543091,
      "learning_rate": 7.854007854007854e-06,
      "loss": 0.0969,
      "step": 1411
    },
    {
      "epoch": 1.4677754677754677,
      "grad_norm": 20.77876853942871,
      "learning_rate": 7.851697851697852e-06,
      "loss": 3.1078,
      "step": 1412
    },
    {
      "epoch": 1.4688149688149688,
      "grad_norm": 7.184508800506592,
      "learning_rate": 7.849387849387849e-06,
      "loss": 0.1976,
      "step": 1413
    },
    {
      "epoch": 1.4698544698544698,
      "grad_norm": 9.331084251403809,
      "learning_rate": 7.847077847077847e-06,
      "loss": 0.3223,
      "step": 1414
    },
    {
      "epoch": 1.4708939708939708,
      "grad_norm": 4.02923583984375,
      "learning_rate": 7.844767844767845e-06,
      "loss": 0.1476,
      "step": 1415
    },
    {
      "epoch": 1.471933471933472,
      "grad_norm": 11.09693717956543,
      "learning_rate": 7.842457842457843e-06,
      "loss": 0.6451,
      "step": 1416
    },
    {
      "epoch": 1.472972972972973,
      "grad_norm": 0.6593981385231018,
      "learning_rate": 7.84014784014784e-06,
      "loss": 0.0162,
      "step": 1417
    },
    {
      "epoch": 1.474012474012474,
      "grad_norm": 7.206985950469971,
      "learning_rate": 7.837837837837838e-06,
      "loss": 0.1954,
      "step": 1418
    },
    {
      "epoch": 1.475051975051975,
      "grad_norm": 0.5606830716133118,
      "learning_rate": 7.835527835527836e-06,
      "loss": 0.008,
      "step": 1419
    },
    {
      "epoch": 1.476091476091476,
      "grad_norm": 17.2354679107666,
      "learning_rate": 7.833217833217834e-06,
      "loss": 1.3854,
      "step": 1420
    },
    {
      "epoch": 1.477130977130977,
      "grad_norm": 0.23821276426315308,
      "learning_rate": 7.830907830907831e-06,
      "loss": 0.0042,
      "step": 1421
    },
    {
      "epoch": 1.4781704781704783,
      "grad_norm": 6.488010406494141,
      "learning_rate": 7.828597828597829e-06,
      "loss": 0.1319,
      "step": 1422
    },
    {
      "epoch": 1.4792099792099793,
      "grad_norm": 20.678817749023438,
      "learning_rate": 7.826287826287827e-06,
      "loss": 0.5714,
      "step": 1423
    },
    {
      "epoch": 1.4802494802494803,
      "grad_norm": 1.193542242050171,
      "learning_rate": 7.823977823977824e-06,
      "loss": 0.025,
      "step": 1424
    },
    {
      "epoch": 1.4812889812889813,
      "grad_norm": 5.369607925415039,
      "learning_rate": 7.821667821667822e-06,
      "loss": 0.1313,
      "step": 1425
    },
    {
      "epoch": 1.4823284823284824,
      "grad_norm": 0.40537557005882263,
      "learning_rate": 7.81935781935782e-06,
      "loss": 0.008,
      "step": 1426
    },
    {
      "epoch": 1.4833679833679834,
      "grad_norm": 13.514300346374512,
      "learning_rate": 7.817047817047818e-06,
      "loss": 0.5955,
      "step": 1427
    },
    {
      "epoch": 1.4844074844074844,
      "grad_norm": 4.4053544998168945,
      "learning_rate": 7.814737814737815e-06,
      "loss": 0.1025,
      "step": 1428
    },
    {
      "epoch": 1.4854469854469854,
      "grad_norm": 2.3163092136383057,
      "learning_rate": 7.812427812427813e-06,
      "loss": 0.0969,
      "step": 1429
    },
    {
      "epoch": 1.4864864864864864,
      "grad_norm": 3.30670428276062,
      "learning_rate": 7.810117810117811e-06,
      "loss": 0.0597,
      "step": 1430
    },
    {
      "epoch": 1.4875259875259874,
      "grad_norm": 13.37387466430664,
      "learning_rate": 7.807807807807808e-06,
      "loss": 1.9395,
      "step": 1431
    },
    {
      "epoch": 1.4885654885654886,
      "grad_norm": 2.1618831157684326,
      "learning_rate": 7.805497805497806e-06,
      "loss": 0.0635,
      "step": 1432
    },
    {
      "epoch": 1.4896049896049897,
      "grad_norm": 0.022710174322128296,
      "learning_rate": 7.803187803187804e-06,
      "loss": 0.0005,
      "step": 1433
    },
    {
      "epoch": 1.4906444906444907,
      "grad_norm": 0.03452178090810776,
      "learning_rate": 7.800877800877802e-06,
      "loss": 0.0007,
      "step": 1434
    },
    {
      "epoch": 1.4916839916839917,
      "grad_norm": 6.462287425994873,
      "learning_rate": 7.798567798567799e-06,
      "loss": 0.3239,
      "step": 1435
    },
    {
      "epoch": 1.4927234927234927,
      "grad_norm": 6.634264945983887,
      "learning_rate": 7.796257796257797e-06,
      "loss": 0.1728,
      "step": 1436
    },
    {
      "epoch": 1.4937629937629937,
      "grad_norm": 9.087811470031738,
      "learning_rate": 7.793947793947795e-06,
      "loss": 0.8339,
      "step": 1437
    },
    {
      "epoch": 1.494802494802495,
      "grad_norm": 4.972245693206787,
      "learning_rate": 7.791637791637793e-06,
      "loss": 0.1508,
      "step": 1438
    },
    {
      "epoch": 1.495841995841996,
      "grad_norm": 9.671388626098633,
      "learning_rate": 7.78932778932779e-06,
      "loss": 0.167,
      "step": 1439
    },
    {
      "epoch": 1.496881496881497,
      "grad_norm": 5.96663236618042,
      "learning_rate": 7.787017787017788e-06,
      "loss": 0.2168,
      "step": 1440
    },
    {
      "epoch": 1.497920997920998,
      "grad_norm": 16.127059936523438,
      "learning_rate": 7.784707784707786e-06,
      "loss": 2.7398,
      "step": 1441
    },
    {
      "epoch": 1.498960498960499,
      "grad_norm": 0.006674590986222029,
      "learning_rate": 7.782397782397783e-06,
      "loss": 0.0001,
      "step": 1442
    },
    {
      "epoch": 1.5,
      "grad_norm": 9.377375602722168,
      "learning_rate": 7.780087780087781e-06,
      "loss": 0.6852,
      "step": 1443
    },
    {
      "epoch": 1.501039501039501,
      "grad_norm": 7.858187675476074,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.405,
      "step": 1444
    },
    {
      "epoch": 1.502079002079002,
      "grad_norm": 7.439113140106201,
      "learning_rate": 7.775467775467777e-06,
      "loss": 0.3658,
      "step": 1445
    },
    {
      "epoch": 1.503118503118503,
      "grad_norm": 6.521304130554199,
      "learning_rate": 7.773157773157774e-06,
      "loss": 0.4594,
      "step": 1446
    },
    {
      "epoch": 1.504158004158004,
      "grad_norm": 13.938838958740234,
      "learning_rate": 7.770847770847772e-06,
      "loss": 0.2794,
      "step": 1447
    },
    {
      "epoch": 1.505197505197505,
      "grad_norm": 8.016144752502441,
      "learning_rate": 7.76853776853777e-06,
      "loss": 0.306,
      "step": 1448
    },
    {
      "epoch": 1.506237006237006,
      "grad_norm": 10.953642845153809,
      "learning_rate": 7.766227766227767e-06,
      "loss": 0.3486,
      "step": 1449
    },
    {
      "epoch": 1.5072765072765073,
      "grad_norm": 0.8176953196525574,
      "learning_rate": 7.763917763917765e-06,
      "loss": 0.0171,
      "step": 1450
    },
    {
      "epoch": 1.5083160083160083,
      "grad_norm": 9.059322357177734,
      "learning_rate": 7.761607761607763e-06,
      "loss": 0.1213,
      "step": 1451
    },
    {
      "epoch": 1.5093555093555093,
      "grad_norm": 3.3097989559173584,
      "learning_rate": 7.75929775929776e-06,
      "loss": 0.1607,
      "step": 1452
    },
    {
      "epoch": 1.5103950103950103,
      "grad_norm": 1.459271788597107,
      "learning_rate": 7.756987756987758e-06,
      "loss": 0.0208,
      "step": 1453
    },
    {
      "epoch": 1.5114345114345116,
      "grad_norm": 2.3589069843292236,
      "learning_rate": 7.754677754677756e-06,
      "loss": 0.0431,
      "step": 1454
    },
    {
      "epoch": 1.5124740124740126,
      "grad_norm": 0.13137201964855194,
      "learning_rate": 7.752367752367752e-06,
      "loss": 0.0036,
      "step": 1455
    },
    {
      "epoch": 1.5135135135135136,
      "grad_norm": 9.798415184020996,
      "learning_rate": 7.75005775005775e-06,
      "loss": 0.3902,
      "step": 1456
    },
    {
      "epoch": 1.5145530145530146,
      "grad_norm": 0.7354431748390198,
      "learning_rate": 7.747747747747749e-06,
      "loss": 0.0218,
      "step": 1457
    },
    {
      "epoch": 1.5155925155925156,
      "grad_norm": 6.521695137023926,
      "learning_rate": 7.745437745437745e-06,
      "loss": 0.2193,
      "step": 1458
    },
    {
      "epoch": 1.5166320166320166,
      "grad_norm": 0.0805174857378006,
      "learning_rate": 7.743127743127744e-06,
      "loss": 0.0023,
      "step": 1459
    },
    {
      "epoch": 1.5176715176715176,
      "grad_norm": 3.0972719192504883,
      "learning_rate": 7.740817740817742e-06,
      "loss": 0.0756,
      "step": 1460
    },
    {
      "epoch": 1.5187110187110187,
      "grad_norm": 9.875079154968262,
      "learning_rate": 7.738507738507738e-06,
      "loss": 0.8773,
      "step": 1461
    },
    {
      "epoch": 1.5197505197505197,
      "grad_norm": 3.0031330585479736,
      "learning_rate": 7.736197736197736e-06,
      "loss": 0.0603,
      "step": 1462
    },
    {
      "epoch": 1.5207900207900207,
      "grad_norm": 14.723185539245605,
      "learning_rate": 7.733887733887735e-06,
      "loss": 1.6017,
      "step": 1463
    },
    {
      "epoch": 1.5218295218295217,
      "grad_norm": 0.344759076833725,
      "learning_rate": 7.731577731577731e-06,
      "loss": 0.0105,
      "step": 1464
    },
    {
      "epoch": 1.5228690228690227,
      "grad_norm": 5.559858322143555,
      "learning_rate": 7.72926772926773e-06,
      "loss": 0.2469,
      "step": 1465
    },
    {
      "epoch": 1.523908523908524,
      "grad_norm": 13.323915481567383,
      "learning_rate": 7.726957726957728e-06,
      "loss": 1.2224,
      "step": 1466
    },
    {
      "epoch": 1.524948024948025,
      "grad_norm": 0.22189731895923615,
      "learning_rate": 7.724647724647726e-06,
      "loss": 0.0073,
      "step": 1467
    },
    {
      "epoch": 1.525987525987526,
      "grad_norm": 0.010859851725399494,
      "learning_rate": 7.722337722337722e-06,
      "loss": 0.0003,
      "step": 1468
    },
    {
      "epoch": 1.527027027027027,
      "grad_norm": 0.7272409796714783,
      "learning_rate": 7.72002772002772e-06,
      "loss": 0.0224,
      "step": 1469
    },
    {
      "epoch": 1.5280665280665282,
      "grad_norm": 5.3687052726745605,
      "learning_rate": 7.717717717717719e-06,
      "loss": 0.4379,
      "step": 1470
    },
    {
      "epoch": 1.5291060291060292,
      "grad_norm": 12.218472480773926,
      "learning_rate": 7.715407715407715e-06,
      "loss": 1.353,
      "step": 1471
    },
    {
      "epoch": 1.5301455301455302,
      "grad_norm": 2.5254576206207275,
      "learning_rate": 7.713097713097713e-06,
      "loss": 0.0473,
      "step": 1472
    },
    {
      "epoch": 1.5311850311850312,
      "grad_norm": 1.6205060482025146,
      "learning_rate": 7.710787710787711e-06,
      "loss": 0.0293,
      "step": 1473
    },
    {
      "epoch": 1.5322245322245323,
      "grad_norm": 7.083399772644043,
      "learning_rate": 7.70847770847771e-06,
      "loss": 0.1893,
      "step": 1474
    },
    {
      "epoch": 1.5332640332640333,
      "grad_norm": 0.48890095949172974,
      "learning_rate": 7.706167706167706e-06,
      "loss": 0.0165,
      "step": 1475
    },
    {
      "epoch": 1.5343035343035343,
      "grad_norm": 13.486111640930176,
      "learning_rate": 7.703857703857704e-06,
      "loss": 0.5817,
      "step": 1476
    },
    {
      "epoch": 1.5353430353430353,
      "grad_norm": 8.082154273986816,
      "learning_rate": 7.701547701547703e-06,
      "loss": 0.1659,
      "step": 1477
    },
    {
      "epoch": 1.5363825363825363,
      "grad_norm": 3.9634504318237305,
      "learning_rate": 7.699237699237699e-06,
      "loss": 0.1039,
      "step": 1478
    },
    {
      "epoch": 1.5374220374220373,
      "grad_norm": 0.2630949318408966,
      "learning_rate": 7.696927696927697e-06,
      "loss": 0.0071,
      "step": 1479
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 8.920835494995117,
      "learning_rate": 7.694617694617695e-06,
      "loss": 0.2141,
      "step": 1480
    },
    {
      "epoch": 1.5395010395010393,
      "grad_norm": 8.480643272399902,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.2279,
      "step": 1481
    },
    {
      "epoch": 1.5405405405405406,
      "grad_norm": 8.147748947143555,
      "learning_rate": 7.68999768999769e-06,
      "loss": 0.2301,
      "step": 1482
    },
    {
      "epoch": 1.5415800415800416,
      "grad_norm": 9.60760498046875,
      "learning_rate": 7.687687687687688e-06,
      "loss": 0.0828,
      "step": 1483
    },
    {
      "epoch": 1.5426195426195426,
      "grad_norm": 0.3811738193035126,
      "learning_rate": 7.685377685377686e-06,
      "loss": 0.0085,
      "step": 1484
    },
    {
      "epoch": 1.5436590436590436,
      "grad_norm": 0.14392603933811188,
      "learning_rate": 7.683067683067685e-06,
      "loss": 0.0029,
      "step": 1485
    },
    {
      "epoch": 1.5446985446985448,
      "grad_norm": 3.819782018661499,
      "learning_rate": 7.680757680757681e-06,
      "loss": 0.2339,
      "step": 1486
    },
    {
      "epoch": 1.5457380457380459,
      "grad_norm": 3.4342427253723145,
      "learning_rate": 7.67844767844768e-06,
      "loss": 0.0857,
      "step": 1487
    },
    {
      "epoch": 1.5467775467775469,
      "grad_norm": 11.615262031555176,
      "learning_rate": 7.676137676137678e-06,
      "loss": 1.1684,
      "step": 1488
    },
    {
      "epoch": 1.5478170478170479,
      "grad_norm": 2.01460599899292,
      "learning_rate": 7.673827673827674e-06,
      "loss": 0.04,
      "step": 1489
    },
    {
      "epoch": 1.5488565488565489,
      "grad_norm": 0.06286811083555222,
      "learning_rate": 7.671517671517672e-06,
      "loss": 0.0011,
      "step": 1490
    },
    {
      "epoch": 1.54989604989605,
      "grad_norm": 12.91536808013916,
      "learning_rate": 7.66920766920767e-06,
      "loss": 0.4833,
      "step": 1491
    },
    {
      "epoch": 1.550935550935551,
      "grad_norm": 0.4687879979610443,
      "learning_rate": 7.666897666897669e-06,
      "loss": 0.0058,
      "step": 1492
    },
    {
      "epoch": 1.551975051975052,
      "grad_norm": 3.517775774002075,
      "learning_rate": 7.664587664587665e-06,
      "loss": 0.1211,
      "step": 1493
    },
    {
      "epoch": 1.553014553014553,
      "grad_norm": 6.1218719482421875,
      "learning_rate": 7.662277662277663e-06,
      "loss": 0.1487,
      "step": 1494
    },
    {
      "epoch": 1.554054054054054,
      "grad_norm": 4.684225559234619,
      "learning_rate": 7.659967659967661e-06,
      "loss": 0.1004,
      "step": 1495
    },
    {
      "epoch": 1.555093555093555,
      "grad_norm": 4.690369606018066,
      "learning_rate": 7.657657657657658e-06,
      "loss": 0.2011,
      "step": 1496
    },
    {
      "epoch": 1.556133056133056,
      "grad_norm": 7.243317127227783,
      "learning_rate": 7.655347655347656e-06,
      "loss": 0.1644,
      "step": 1497
    },
    {
      "epoch": 1.5571725571725572,
      "grad_norm": 4.194143772125244,
      "learning_rate": 7.653037653037654e-06,
      "loss": 0.1912,
      "step": 1498
    },
    {
      "epoch": 1.5582120582120582,
      "grad_norm": 3.54541277885437,
      "learning_rate": 7.650727650727653e-06,
      "loss": 0.0766,
      "step": 1499
    },
    {
      "epoch": 1.5592515592515592,
      "grad_norm": 5.809798240661621,
      "learning_rate": 7.648417648417649e-06,
      "loss": 0.1385,
      "step": 1500
    },
    {
      "epoch": 1.5602910602910602,
      "grad_norm": 0.055198103189468384,
      "learning_rate": 7.646107646107647e-06,
      "loss": 0.0016,
      "step": 1501
    },
    {
      "epoch": 1.5613305613305615,
      "grad_norm": 0.09694336354732513,
      "learning_rate": 7.643797643797645e-06,
      "loss": 0.0024,
      "step": 1502
    },
    {
      "epoch": 1.5623700623700625,
      "grad_norm": 6.44903564453125,
      "learning_rate": 7.641487641487642e-06,
      "loss": 0.3952,
      "step": 1503
    },
    {
      "epoch": 1.5634095634095635,
      "grad_norm": 0.7087557315826416,
      "learning_rate": 7.63917763917764e-06,
      "loss": 0.0143,
      "step": 1504
    },
    {
      "epoch": 1.5644490644490645,
      "grad_norm": 8.263898849487305,
      "learning_rate": 7.636867636867638e-06,
      "loss": 0.5762,
      "step": 1505
    },
    {
      "epoch": 1.5654885654885655,
      "grad_norm": 2.20043683052063,
      "learning_rate": 7.634557634557635e-06,
      "loss": 0.0495,
      "step": 1506
    },
    {
      "epoch": 1.5665280665280665,
      "grad_norm": 4.385960578918457,
      "learning_rate": 7.632247632247633e-06,
      "loss": 0.0778,
      "step": 1507
    },
    {
      "epoch": 1.5675675675675675,
      "grad_norm": 0.7370150685310364,
      "learning_rate": 7.629937629937631e-06,
      "loss": 0.0202,
      "step": 1508
    },
    {
      "epoch": 1.5686070686070686,
      "grad_norm": 0.4597228467464447,
      "learning_rate": 7.6276276276276285e-06,
      "loss": 0.0069,
      "step": 1509
    },
    {
      "epoch": 1.5696465696465696,
      "grad_norm": 4.463411331176758,
      "learning_rate": 7.625317625317627e-06,
      "loss": 0.0576,
      "step": 1510
    },
    {
      "epoch": 1.5706860706860706,
      "grad_norm": 10.788517951965332,
      "learning_rate": 7.623007623007623e-06,
      "loss": 0.7124,
      "step": 1511
    },
    {
      "epoch": 1.5717255717255716,
      "grad_norm": 5.738900184631348,
      "learning_rate": 7.620697620697621e-06,
      "loss": 0.2167,
      "step": 1512
    },
    {
      "epoch": 1.5727650727650726,
      "grad_norm": 10.37417221069336,
      "learning_rate": 7.61838761838762e-06,
      "loss": 0.4828,
      "step": 1513
    },
    {
      "epoch": 1.5738045738045738,
      "grad_norm": 8.865348815917969,
      "learning_rate": 7.616077616077616e-06,
      "loss": 0.7257,
      "step": 1514
    },
    {
      "epoch": 1.5748440748440748,
      "grad_norm": 1.017466425895691,
      "learning_rate": 7.613767613767614e-06,
      "loss": 0.0313,
      "step": 1515
    },
    {
      "epoch": 1.5758835758835759,
      "grad_norm": 0.827038586139679,
      "learning_rate": 7.6114576114576125e-06,
      "loss": 0.028,
      "step": 1516
    },
    {
      "epoch": 1.5769230769230769,
      "grad_norm": 8.467267990112305,
      "learning_rate": 7.60914760914761e-06,
      "loss": 0.7333,
      "step": 1517
    },
    {
      "epoch": 1.577962577962578,
      "grad_norm": 4.1905646324157715,
      "learning_rate": 7.606837606837607e-06,
      "loss": 0.1714,
      "step": 1518
    },
    {
      "epoch": 1.5790020790020791,
      "grad_norm": 1.6618033647537231,
      "learning_rate": 7.604527604527605e-06,
      "loss": 0.0323,
      "step": 1519
    },
    {
      "epoch": 1.5800415800415801,
      "grad_norm": 2.4269134998321533,
      "learning_rate": 7.602217602217603e-06,
      "loss": 0.0614,
      "step": 1520
    },
    {
      "epoch": 1.5810810810810811,
      "grad_norm": 10.702552795410156,
      "learning_rate": 7.599907599907601e-06,
      "loss": 0.8274,
      "step": 1521
    },
    {
      "epoch": 1.5821205821205822,
      "grad_norm": 9.825342178344727,
      "learning_rate": 7.597597597597598e-06,
      "loss": 2.324,
      "step": 1522
    },
    {
      "epoch": 1.5831600831600832,
      "grad_norm": 2.365910291671753,
      "learning_rate": 7.5952875952875956e-06,
      "loss": 0.0784,
      "step": 1523
    },
    {
      "epoch": 1.5841995841995842,
      "grad_norm": 0.04099468141794205,
      "learning_rate": 7.592977592977594e-06,
      "loss": 0.0007,
      "step": 1524
    },
    {
      "epoch": 1.5852390852390852,
      "grad_norm": 11.147265434265137,
      "learning_rate": 7.590667590667591e-06,
      "loss": 0.3298,
      "step": 1525
    },
    {
      "epoch": 1.5862785862785862,
      "grad_norm": 0.05397927388548851,
      "learning_rate": 7.5883575883575885e-06,
      "loss": 0.001,
      "step": 1526
    },
    {
      "epoch": 1.5873180873180872,
      "grad_norm": 2.043637990951538,
      "learning_rate": 7.586047586047587e-06,
      "loss": 0.0426,
      "step": 1527
    },
    {
      "epoch": 1.5883575883575882,
      "grad_norm": 0.0009239251958206296,
      "learning_rate": 7.583737583737585e-06,
      "loss": 0.0,
      "step": 1528
    },
    {
      "epoch": 1.5893970893970892,
      "grad_norm": 0.13833796977996826,
      "learning_rate": 7.581427581427581e-06,
      "loss": 0.0022,
      "step": 1529
    },
    {
      "epoch": 1.5904365904365905,
      "grad_norm": 9.737223625183105,
      "learning_rate": 7.5791175791175795e-06,
      "loss": 0.4196,
      "step": 1530
    },
    {
      "epoch": 1.5914760914760915,
      "grad_norm": 4.808546543121338,
      "learning_rate": 7.576807576807578e-06,
      "loss": 0.1391,
      "step": 1531
    },
    {
      "epoch": 1.5925155925155925,
      "grad_norm": 7.235491752624512,
      "learning_rate": 7.574497574497574e-06,
      "loss": 0.1397,
      "step": 1532
    },
    {
      "epoch": 1.5935550935550935,
      "grad_norm": 0.9746120572090149,
      "learning_rate": 7.572187572187572e-06,
      "loss": 0.0173,
      "step": 1533
    },
    {
      "epoch": 1.5945945945945947,
      "grad_norm": 14.863147735595703,
      "learning_rate": 7.569877569877571e-06,
      "loss": 1.9486,
      "step": 1534
    },
    {
      "epoch": 1.5956340956340958,
      "grad_norm": 11.282873153686523,
      "learning_rate": 7.567567567567569e-06,
      "loss": 0.6396,
      "step": 1535
    },
    {
      "epoch": 1.5966735966735968,
      "grad_norm": 0.25318750739097595,
      "learning_rate": 7.565257565257565e-06,
      "loss": 0.0038,
      "step": 1536
    },
    {
      "epoch": 1.5977130977130978,
      "grad_norm": 1.9883997440338135,
      "learning_rate": 7.5629475629475635e-06,
      "loss": 0.0468,
      "step": 1537
    },
    {
      "epoch": 1.5987525987525988,
      "grad_norm": 6.504814147949219,
      "learning_rate": 7.560637560637562e-06,
      "loss": 0.3667,
      "step": 1538
    },
    {
      "epoch": 1.5997920997920998,
      "grad_norm": 2.2447383403778076,
      "learning_rate": 7.55832755832756e-06,
      "loss": 0.0298,
      "step": 1539
    },
    {
      "epoch": 1.6008316008316008,
      "grad_norm": 0.31640592217445374,
      "learning_rate": 7.556017556017556e-06,
      "loss": 0.0111,
      "step": 1540
    },
    {
      "epoch": 1.6018711018711018,
      "grad_norm": 7.6291399002075195,
      "learning_rate": 7.5537075537075545e-06,
      "loss": 0.1263,
      "step": 1541
    },
    {
      "epoch": 1.6029106029106028,
      "grad_norm": 3.217961549758911,
      "learning_rate": 7.551397551397553e-06,
      "loss": 0.1511,
      "step": 1542
    },
    {
      "epoch": 1.6039501039501038,
      "grad_norm": 2.8815371990203857,
      "learning_rate": 7.549087549087549e-06,
      "loss": 0.0455,
      "step": 1543
    },
    {
      "epoch": 1.6049896049896049,
      "grad_norm": 1.2733142375946045,
      "learning_rate": 7.546777546777547e-06,
      "loss": 0.0268,
      "step": 1544
    },
    {
      "epoch": 1.6060291060291059,
      "grad_norm": 0.5411023497581482,
      "learning_rate": 7.544467544467546e-06,
      "loss": 0.0096,
      "step": 1545
    },
    {
      "epoch": 1.607068607068607,
      "grad_norm": 4.737339496612549,
      "learning_rate": 7.542157542157543e-06,
      "loss": 0.1374,
      "step": 1546
    },
    {
      "epoch": 1.6081081081081081,
      "grad_norm": 0.9114570021629333,
      "learning_rate": 7.53984753984754e-06,
      "loss": 0.0267,
      "step": 1547
    },
    {
      "epoch": 1.6091476091476091,
      "grad_norm": 0.3017042279243469,
      "learning_rate": 7.5375375375375385e-06,
      "loss": 0.0069,
      "step": 1548
    },
    {
      "epoch": 1.6101871101871101,
      "grad_norm": 8.151854515075684,
      "learning_rate": 7.535227535227536e-06,
      "loss": 0.2909,
      "step": 1549
    },
    {
      "epoch": 1.6112266112266114,
      "grad_norm": 0.5599004626274109,
      "learning_rate": 7.532917532917533e-06,
      "loss": 0.0167,
      "step": 1550
    },
    {
      "epoch": 1.6122661122661124,
      "grad_norm": 4.013564109802246,
      "learning_rate": 7.530607530607531e-06,
      "loss": 0.1025,
      "step": 1551
    },
    {
      "epoch": 1.6133056133056134,
      "grad_norm": 6.159677028656006,
      "learning_rate": 7.528297528297529e-06,
      "loss": 0.087,
      "step": 1552
    },
    {
      "epoch": 1.6143451143451144,
      "grad_norm": 0.3326902389526367,
      "learning_rate": 7.525987525987527e-06,
      "loss": 0.0077,
      "step": 1553
    },
    {
      "epoch": 1.6153846153846154,
      "grad_norm": 13.508583068847656,
      "learning_rate": 7.523677523677524e-06,
      "loss": 1.1188,
      "step": 1554
    },
    {
      "epoch": 1.6164241164241164,
      "grad_norm": 9.111117362976074,
      "learning_rate": 7.521367521367522e-06,
      "loss": 0.177,
      "step": 1555
    },
    {
      "epoch": 1.6174636174636174,
      "grad_norm": 0.7300674319267273,
      "learning_rate": 7.51905751905752e-06,
      "loss": 0.0162,
      "step": 1556
    },
    {
      "epoch": 1.6185031185031185,
      "grad_norm": 6.406063556671143,
      "learning_rate": 7.516747516747518e-06,
      "loss": 0.1966,
      "step": 1557
    },
    {
      "epoch": 1.6195426195426195,
      "grad_norm": 12.330337524414062,
      "learning_rate": 7.5144375144375145e-06,
      "loss": 2.1292,
      "step": 1558
    },
    {
      "epoch": 1.6205821205821205,
      "grad_norm": 1.7357144355773926,
      "learning_rate": 7.512127512127513e-06,
      "loss": 0.0297,
      "step": 1559
    },
    {
      "epoch": 1.6216216216216215,
      "grad_norm": 3.0343379974365234,
      "learning_rate": 7.509817509817511e-06,
      "loss": 0.0694,
      "step": 1560
    },
    {
      "epoch": 1.6226611226611225,
      "grad_norm": 0.01201736181974411,
      "learning_rate": 7.507507507507507e-06,
      "loss": 0.0003,
      "step": 1561
    },
    {
      "epoch": 1.6237006237006237,
      "grad_norm": 21.220945358276367,
      "learning_rate": 7.5051975051975055e-06,
      "loss": 2.9917,
      "step": 1562
    },
    {
      "epoch": 1.6247401247401247,
      "grad_norm": 6.297361850738525,
      "learning_rate": 7.502887502887504e-06,
      "loss": 0.1415,
      "step": 1563
    },
    {
      "epoch": 1.6257796257796258,
      "grad_norm": 8.065841674804688,
      "learning_rate": 7.500577500577502e-06,
      "loss": 0.3198,
      "step": 1564
    },
    {
      "epoch": 1.6268191268191268,
      "grad_norm": 6.228341579437256,
      "learning_rate": 7.498267498267498e-06,
      "loss": 0.3259,
      "step": 1565
    },
    {
      "epoch": 1.627858627858628,
      "grad_norm": 5.884280204772949,
      "learning_rate": 7.495957495957497e-06,
      "loss": 0.2876,
      "step": 1566
    },
    {
      "epoch": 1.628898128898129,
      "grad_norm": 13.760956764221191,
      "learning_rate": 7.493647493647495e-06,
      "loss": 0.885,
      "step": 1567
    },
    {
      "epoch": 1.62993762993763,
      "grad_norm": 11.207575798034668,
      "learning_rate": 7.491337491337491e-06,
      "loss": 0.8122,
      "step": 1568
    },
    {
      "epoch": 1.630977130977131,
      "grad_norm": 0.14747324585914612,
      "learning_rate": 7.4890274890274895e-06,
      "loss": 0.0024,
      "step": 1569
    },
    {
      "epoch": 1.632016632016632,
      "grad_norm": 6.728248119354248,
      "learning_rate": 7.486717486717488e-06,
      "loss": 0.2973,
      "step": 1570
    },
    {
      "epoch": 1.633056133056133,
      "grad_norm": 0.09715936332941055,
      "learning_rate": 7.484407484407485e-06,
      "loss": 0.0018,
      "step": 1571
    },
    {
      "epoch": 1.634095634095634,
      "grad_norm": 2.335740804672241,
      "learning_rate": 7.482097482097482e-06,
      "loss": 0.0711,
      "step": 1572
    },
    {
      "epoch": 1.635135135135135,
      "grad_norm": 2.9282028675079346,
      "learning_rate": 7.4797874797874805e-06,
      "loss": 0.058,
      "step": 1573
    },
    {
      "epoch": 1.636174636174636,
      "grad_norm": 2.8667209148406982,
      "learning_rate": 7.477477477477479e-06,
      "loss": 0.0595,
      "step": 1574
    },
    {
      "epoch": 1.637214137214137,
      "grad_norm": 7.022189617156982,
      "learning_rate": 7.475167475167476e-06,
      "loss": 0.3922,
      "step": 1575
    },
    {
      "epoch": 1.6382536382536381,
      "grad_norm": 8.579756736755371,
      "learning_rate": 7.4728574728574734e-06,
      "loss": 0.208,
      "step": 1576
    },
    {
      "epoch": 1.6392931392931391,
      "grad_norm": 12.531072616577148,
      "learning_rate": 7.470547470547472e-06,
      "loss": 0.844,
      "step": 1577
    },
    {
      "epoch": 1.6403326403326404,
      "grad_norm": 2.289039373397827,
      "learning_rate": 7.468237468237469e-06,
      "loss": 0.0383,
      "step": 1578
    },
    {
      "epoch": 1.6413721413721414,
      "grad_norm": 11.637489318847656,
      "learning_rate": 7.465927465927466e-06,
      "loss": 0.4017,
      "step": 1579
    },
    {
      "epoch": 1.6424116424116424,
      "grad_norm": 0.017794983461499214,
      "learning_rate": 7.4636174636174645e-06,
      "loss": 0.0005,
      "step": 1580
    },
    {
      "epoch": 1.6434511434511434,
      "grad_norm": 10.051448822021484,
      "learning_rate": 7.461307461307462e-06,
      "loss": 1.1899,
      "step": 1581
    },
    {
      "epoch": 1.6444906444906446,
      "grad_norm": 0.20261698961257935,
      "learning_rate": 7.45899745899746e-06,
      "loss": 0.0038,
      "step": 1582
    },
    {
      "epoch": 1.6455301455301456,
      "grad_norm": 2.199592113494873,
      "learning_rate": 7.456687456687457e-06,
      "loss": 0.0379,
      "step": 1583
    },
    {
      "epoch": 1.6465696465696467,
      "grad_norm": 0.1683259755373001,
      "learning_rate": 7.454377454377455e-06,
      "loss": 0.0032,
      "step": 1584
    },
    {
      "epoch": 1.6476091476091477,
      "grad_norm": 6.000250339508057,
      "learning_rate": 7.452067452067453e-06,
      "loss": 0.2425,
      "step": 1585
    },
    {
      "epoch": 1.6486486486486487,
      "grad_norm": 0.08056347072124481,
      "learning_rate": 7.44975744975745e-06,
      "loss": 0.0025,
      "step": 1586
    },
    {
      "epoch": 1.6496881496881497,
      "grad_norm": 9.408105850219727,
      "learning_rate": 7.447447447447448e-06,
      "loss": 0.3774,
      "step": 1587
    },
    {
      "epoch": 1.6507276507276507,
      "grad_norm": 6.828577995300293,
      "learning_rate": 7.445137445137446e-06,
      "loss": 0.5208,
      "step": 1588
    },
    {
      "epoch": 1.6517671517671517,
      "grad_norm": 9.918600082397461,
      "learning_rate": 7.442827442827444e-06,
      "loss": 0.6716,
      "step": 1589
    },
    {
      "epoch": 1.6528066528066527,
      "grad_norm": 1.9643577337265015,
      "learning_rate": 7.4405174405174405e-06,
      "loss": 0.0253,
      "step": 1590
    },
    {
      "epoch": 1.6538461538461537,
      "grad_norm": 0.17032547295093536,
      "learning_rate": 7.438207438207439e-06,
      "loss": 0.0032,
      "step": 1591
    },
    {
      "epoch": 1.6548856548856548,
      "grad_norm": 9.601398468017578,
      "learning_rate": 7.435897435897437e-06,
      "loss": 0.1225,
      "step": 1592
    },
    {
      "epoch": 1.6559251559251558,
      "grad_norm": 0.04000294581055641,
      "learning_rate": 7.433587433587435e-06,
      "loss": 0.0014,
      "step": 1593
    },
    {
      "epoch": 1.656964656964657,
      "grad_norm": 1.3193373680114746,
      "learning_rate": 7.4312774312774315e-06,
      "loss": 0.0365,
      "step": 1594
    },
    {
      "epoch": 1.658004158004158,
      "grad_norm": 1.8269625902175903,
      "learning_rate": 7.42896742896743e-06,
      "loss": 0.0426,
      "step": 1595
    },
    {
      "epoch": 1.659043659043659,
      "grad_norm": 0.24532830715179443,
      "learning_rate": 7.426657426657428e-06,
      "loss": 0.0055,
      "step": 1596
    },
    {
      "epoch": 1.66008316008316,
      "grad_norm": 1.368507981300354,
      "learning_rate": 7.424347424347424e-06,
      "loss": 0.0634,
      "step": 1597
    },
    {
      "epoch": 1.6611226611226613,
      "grad_norm": 0.16978596150875092,
      "learning_rate": 7.422037422037423e-06,
      "loss": 0.0031,
      "step": 1598
    },
    {
      "epoch": 1.6621621621621623,
      "grad_norm": 2.2428948879241943,
      "learning_rate": 7.419727419727421e-06,
      "loss": 0.0447,
      "step": 1599
    },
    {
      "epoch": 1.6632016632016633,
      "grad_norm": 1.2490960359573364,
      "learning_rate": 7.417417417417418e-06,
      "loss": 0.0241,
      "step": 1600
    },
    {
      "epoch": 1.6642411642411643,
      "grad_norm": 5.034488677978516,
      "learning_rate": 7.4151074151074155e-06,
      "loss": 0.3478,
      "step": 1601
    },
    {
      "epoch": 1.6652806652806653,
      "grad_norm": 0.5424055457115173,
      "learning_rate": 7.412797412797414e-06,
      "loss": 0.0111,
      "step": 1602
    },
    {
      "epoch": 1.6663201663201663,
      "grad_norm": 0.16386502981185913,
      "learning_rate": 7.410487410487411e-06,
      "loss": 0.003,
      "step": 1603
    },
    {
      "epoch": 1.6673596673596673,
      "grad_norm": 12.852590560913086,
      "learning_rate": 7.408177408177408e-06,
      "loss": 0.6814,
      "step": 1604
    },
    {
      "epoch": 1.6683991683991684,
      "grad_norm": 3.859780788421631,
      "learning_rate": 7.4058674058674066e-06,
      "loss": 0.0922,
      "step": 1605
    },
    {
      "epoch": 1.6694386694386694,
      "grad_norm": 3.8476450443267822,
      "learning_rate": 7.403557403557404e-06,
      "loss": 0.1143,
      "step": 1606
    },
    {
      "epoch": 1.6704781704781704,
      "grad_norm": 14.699956893920898,
      "learning_rate": 7.401247401247402e-06,
      "loss": 0.1259,
      "step": 1607
    },
    {
      "epoch": 1.6715176715176714,
      "grad_norm": 0.9650105834007263,
      "learning_rate": 7.3989373989373994e-06,
      "loss": 0.0456,
      "step": 1608
    },
    {
      "epoch": 1.6725571725571724,
      "grad_norm": 0.15530744194984436,
      "learning_rate": 7.396627396627397e-06,
      "loss": 0.0047,
      "step": 1609
    },
    {
      "epoch": 1.6735966735966736,
      "grad_norm": 4.704723358154297,
      "learning_rate": 7.394317394317395e-06,
      "loss": 0.2454,
      "step": 1610
    },
    {
      "epoch": 1.6746361746361746,
      "grad_norm": 4.737837314605713,
      "learning_rate": 7.392007392007393e-06,
      "loss": 0.1039,
      "step": 1611
    },
    {
      "epoch": 1.6756756756756757,
      "grad_norm": 8.195130348205566,
      "learning_rate": 7.38969738969739e-06,
      "loss": 0.419,
      "step": 1612
    },
    {
      "epoch": 1.6767151767151767,
      "grad_norm": 0.1998622566461563,
      "learning_rate": 7.387387387387388e-06,
      "loss": 0.004,
      "step": 1613
    },
    {
      "epoch": 1.677754677754678,
      "grad_norm": 2.4961326122283936,
      "learning_rate": 7.385077385077386e-06,
      "loss": 0.117,
      "step": 1614
    },
    {
      "epoch": 1.678794178794179,
      "grad_norm": 5.578335285186768,
      "learning_rate": 7.3827673827673825e-06,
      "loss": 0.1941,
      "step": 1615
    },
    {
      "epoch": 1.67983367983368,
      "grad_norm": 7.483165264129639,
      "learning_rate": 7.380457380457381e-06,
      "loss": 0.7208,
      "step": 1616
    },
    {
      "epoch": 1.680873180873181,
      "grad_norm": 2.0960891246795654,
      "learning_rate": 7.378147378147379e-06,
      "loss": 0.0293,
      "step": 1617
    },
    {
      "epoch": 1.681912681912682,
      "grad_norm": 7.603874206542969,
      "learning_rate": 7.375837375837377e-06,
      "loss": 0.2035,
      "step": 1618
    },
    {
      "epoch": 1.682952182952183,
      "grad_norm": 0.9122418165206909,
      "learning_rate": 7.373527373527374e-06,
      "loss": 0.0224,
      "step": 1619
    },
    {
      "epoch": 1.683991683991684,
      "grad_norm": 6.815097332000732,
      "learning_rate": 7.371217371217372e-06,
      "loss": 0.1871,
      "step": 1620
    },
    {
      "epoch": 1.685031185031185,
      "grad_norm": 11.70824909210205,
      "learning_rate": 7.36890736890737e-06,
      "loss": 0.7655,
      "step": 1621
    },
    {
      "epoch": 1.686070686070686,
      "grad_norm": 0.46677353978157043,
      "learning_rate": 7.3665973665973665e-06,
      "loss": 0.008,
      "step": 1622
    },
    {
      "epoch": 1.687110187110187,
      "grad_norm": 2.1971335411071777,
      "learning_rate": 7.364287364287365e-06,
      "loss": 0.0468,
      "step": 1623
    },
    {
      "epoch": 1.688149688149688,
      "grad_norm": 5.790288925170898,
      "learning_rate": 7.361977361977363e-06,
      "loss": 0.2241,
      "step": 1624
    },
    {
      "epoch": 1.689189189189189,
      "grad_norm": 4.230993747711182,
      "learning_rate": 7.359667359667361e-06,
      "loss": 0.1058,
      "step": 1625
    },
    {
      "epoch": 1.6902286902286903,
      "grad_norm": 1.344787359237671,
      "learning_rate": 7.3573573573573575e-06,
      "loss": 0.0207,
      "step": 1626
    },
    {
      "epoch": 1.6912681912681913,
      "grad_norm": 2.8183813095092773,
      "learning_rate": 7.355047355047356e-06,
      "loss": 0.067,
      "step": 1627
    },
    {
      "epoch": 1.6923076923076923,
      "grad_norm": 6.174064636230469,
      "learning_rate": 7.352737352737354e-06,
      "loss": 0.2275,
      "step": 1628
    },
    {
      "epoch": 1.6933471933471933,
      "grad_norm": 5.060527801513672,
      "learning_rate": 7.350427350427351e-06,
      "loss": 0.2279,
      "step": 1629
    },
    {
      "epoch": 1.6943866943866945,
      "grad_norm": 5.803682327270508,
      "learning_rate": 7.348117348117349e-06,
      "loss": 0.0884,
      "step": 1630
    },
    {
      "epoch": 1.6954261954261955,
      "grad_norm": 2.137218713760376,
      "learning_rate": 7.345807345807347e-06,
      "loss": 0.1781,
      "step": 1631
    },
    {
      "epoch": 1.6964656964656966,
      "grad_norm": 9.369726181030273,
      "learning_rate": 7.343497343497344e-06,
      "loss": 0.2808,
      "step": 1632
    },
    {
      "epoch": 1.6975051975051976,
      "grad_norm": 3.2292776107788086,
      "learning_rate": 7.3411873411873415e-06,
      "loss": 0.054,
      "step": 1633
    },
    {
      "epoch": 1.6985446985446986,
      "grad_norm": 8.865103721618652,
      "learning_rate": 7.33887733887734e-06,
      "loss": 0.2376,
      "step": 1634
    },
    {
      "epoch": 1.6995841995841996,
      "grad_norm": 4.944544315338135,
      "learning_rate": 7.336567336567337e-06,
      "loss": 0.1071,
      "step": 1635
    },
    {
      "epoch": 1.7006237006237006,
      "grad_norm": 3.2673540115356445,
      "learning_rate": 7.334257334257335e-06,
      "loss": 0.0807,
      "step": 1636
    },
    {
      "epoch": 1.7016632016632016,
      "grad_norm": 8.087736129760742,
      "learning_rate": 7.3319473319473326e-06,
      "loss": 0.3968,
      "step": 1637
    },
    {
      "epoch": 1.7027027027027026,
      "grad_norm": 10.094491958618164,
      "learning_rate": 7.32963732963733e-06,
      "loss": 0.2602,
      "step": 1638
    },
    {
      "epoch": 1.7037422037422036,
      "grad_norm": 5.964760780334473,
      "learning_rate": 7.327327327327328e-06,
      "loss": 0.1667,
      "step": 1639
    },
    {
      "epoch": 1.7047817047817047,
      "grad_norm": 8.874397277832031,
      "learning_rate": 7.3250173250173254e-06,
      "loss": 0.4751,
      "step": 1640
    },
    {
      "epoch": 1.7058212058212057,
      "grad_norm": 14.104971885681152,
      "learning_rate": 7.322707322707323e-06,
      "loss": 1.1987,
      "step": 1641
    },
    {
      "epoch": 1.706860706860707,
      "grad_norm": 14.051151275634766,
      "learning_rate": 7.320397320397321e-06,
      "loss": 1.1578,
      "step": 1642
    },
    {
      "epoch": 1.707900207900208,
      "grad_norm": 0.019640829414129257,
      "learning_rate": 7.318087318087319e-06,
      "loss": 0.0006,
      "step": 1643
    },
    {
      "epoch": 1.708939708939709,
      "grad_norm": 11.411649703979492,
      "learning_rate": 7.315777315777316e-06,
      "loss": 1.4277,
      "step": 1644
    },
    {
      "epoch": 1.70997920997921,
      "grad_norm": 0.04116674140095711,
      "learning_rate": 7.313467313467314e-06,
      "loss": 0.0009,
      "step": 1645
    },
    {
      "epoch": 1.7110187110187112,
      "grad_norm": 0.9147244095802307,
      "learning_rate": 7.311157311157312e-06,
      "loss": 0.0226,
      "step": 1646
    },
    {
      "epoch": 1.7120582120582122,
      "grad_norm": 10.076512336730957,
      "learning_rate": 7.30884730884731e-06,
      "loss": 0.7167,
      "step": 1647
    },
    {
      "epoch": 1.7130977130977132,
      "grad_norm": 7.493573188781738,
      "learning_rate": 7.306537306537307e-06,
      "loss": 0.2172,
      "step": 1648
    },
    {
      "epoch": 1.7141372141372142,
      "grad_norm": 5.900646209716797,
      "learning_rate": 7.304227304227305e-06,
      "loss": 0.0806,
      "step": 1649
    },
    {
      "epoch": 1.7151767151767152,
      "grad_norm": 3.126075029373169,
      "learning_rate": 7.301917301917303e-06,
      "loss": 0.247,
      "step": 1650
    },
    {
      "epoch": 1.7162162162162162,
      "grad_norm": 6.303007125854492,
      "learning_rate": 7.2996072996073e-06,
      "loss": 0.2534,
      "step": 1651
    },
    {
      "epoch": 1.7172557172557172,
      "grad_norm": 11.120226860046387,
      "learning_rate": 7.297297297297298e-06,
      "loss": 0.6595,
      "step": 1652
    },
    {
      "epoch": 1.7182952182952183,
      "grad_norm": 0.505618155002594,
      "learning_rate": 7.294987294987296e-06,
      "loss": 0.0075,
      "step": 1653
    },
    {
      "epoch": 1.7193347193347193,
      "grad_norm": 8.373292922973633,
      "learning_rate": 7.292677292677293e-06,
      "loss": 0.6847,
      "step": 1654
    },
    {
      "epoch": 1.7203742203742203,
      "grad_norm": 6.690762996673584,
      "learning_rate": 7.290367290367291e-06,
      "loss": 0.2232,
      "step": 1655
    },
    {
      "epoch": 1.7214137214137213,
      "grad_norm": 0.13206851482391357,
      "learning_rate": 7.288057288057289e-06,
      "loss": 0.0024,
      "step": 1656
    },
    {
      "epoch": 1.7224532224532223,
      "grad_norm": 7.297382831573486,
      "learning_rate": 7.285747285747286e-06,
      "loss": 0.4353,
      "step": 1657
    },
    {
      "epoch": 1.7234927234927235,
      "grad_norm": 1.3188226222991943,
      "learning_rate": 7.2834372834372836e-06,
      "loss": 0.0408,
      "step": 1658
    },
    {
      "epoch": 1.7245322245322245,
      "grad_norm": 7.577664852142334,
      "learning_rate": 7.281127281127282e-06,
      "loss": 0.2984,
      "step": 1659
    },
    {
      "epoch": 1.7255717255717256,
      "grad_norm": 1.5589632987976074,
      "learning_rate": 7.278817278817279e-06,
      "loss": 0.0465,
      "step": 1660
    },
    {
      "epoch": 1.7266112266112266,
      "grad_norm": 7.908683776855469,
      "learning_rate": 7.276507276507277e-06,
      "loss": 0.7164,
      "step": 1661
    },
    {
      "epoch": 1.7276507276507278,
      "grad_norm": 11.915677070617676,
      "learning_rate": 7.274197274197275e-06,
      "loss": 0.8969,
      "step": 1662
    },
    {
      "epoch": 1.7286902286902288,
      "grad_norm": 8.111615180969238,
      "learning_rate": 7.271887271887272e-06,
      "loss": 0.3777,
      "step": 1663
    },
    {
      "epoch": 1.7297297297297298,
      "grad_norm": 0.46552711725234985,
      "learning_rate": 7.26957726957727e-06,
      "loss": 0.0113,
      "step": 1664
    },
    {
      "epoch": 1.7307692307692308,
      "grad_norm": 4.33073616027832,
      "learning_rate": 7.267267267267268e-06,
      "loss": 0.2095,
      "step": 1665
    },
    {
      "epoch": 1.7318087318087318,
      "grad_norm": 17.092309951782227,
      "learning_rate": 7.264957264957266e-06,
      "loss": 0.7215,
      "step": 1666
    },
    {
      "epoch": 1.7328482328482329,
      "grad_norm": 1.5251858234405518,
      "learning_rate": 7.262647262647263e-06,
      "loss": 0.0386,
      "step": 1667
    },
    {
      "epoch": 1.7338877338877339,
      "grad_norm": 0.3032921254634857,
      "learning_rate": 7.260337260337261e-06,
      "loss": 0.0099,
      "step": 1668
    },
    {
      "epoch": 1.7349272349272349,
      "grad_norm": 11.229517936706543,
      "learning_rate": 7.2580272580272586e-06,
      "loss": 0.7061,
      "step": 1669
    },
    {
      "epoch": 1.735966735966736,
      "grad_norm": 6.576710224151611,
      "learning_rate": 7.255717255717256e-06,
      "loss": 0.3782,
      "step": 1670
    },
    {
      "epoch": 1.737006237006237,
      "grad_norm": 1.0471800565719604,
      "learning_rate": 7.253407253407254e-06,
      "loss": 0.0211,
      "step": 1671
    },
    {
      "epoch": 1.738045738045738,
      "grad_norm": 4.3084306716918945,
      "learning_rate": 7.251097251097252e-06,
      "loss": 0.2368,
      "step": 1672
    },
    {
      "epoch": 1.739085239085239,
      "grad_norm": 1.829033613204956,
      "learning_rate": 7.248787248787249e-06,
      "loss": 0.0723,
      "step": 1673
    },
    {
      "epoch": 1.7401247401247402,
      "grad_norm": 6.535826683044434,
      "learning_rate": 7.246477246477247e-06,
      "loss": 0.4326,
      "step": 1674
    },
    {
      "epoch": 1.7411642411642412,
      "grad_norm": 0.2173418551683426,
      "learning_rate": 7.244167244167245e-06,
      "loss": 0.0058,
      "step": 1675
    },
    {
      "epoch": 1.7422037422037422,
      "grad_norm": 1.3365967273712158,
      "learning_rate": 7.241857241857242e-06,
      "loss": 0.0298,
      "step": 1676
    },
    {
      "epoch": 1.7432432432432432,
      "grad_norm": 0.48751041293144226,
      "learning_rate": 7.23954723954724e-06,
      "loss": 0.0104,
      "step": 1677
    },
    {
      "epoch": 1.7442827442827444,
      "grad_norm": 0.29479509592056274,
      "learning_rate": 7.237237237237238e-06,
      "loss": 0.0054,
      "step": 1678
    },
    {
      "epoch": 1.7453222453222454,
      "grad_norm": 0.020942412316799164,
      "learning_rate": 7.234927234927236e-06,
      "loss": 0.0007,
      "step": 1679
    },
    {
      "epoch": 1.7463617463617465,
      "grad_norm": 13.055707931518555,
      "learning_rate": 7.232617232617233e-06,
      "loss": 0.7853,
      "step": 1680
    },
    {
      "epoch": 1.7474012474012475,
      "grad_norm": 0.0012717257486656308,
      "learning_rate": 7.230307230307231e-06,
      "loss": 0.0,
      "step": 1681
    },
    {
      "epoch": 1.7484407484407485,
      "grad_norm": 0.08711827546358109,
      "learning_rate": 7.227997227997229e-06,
      "loss": 0.0018,
      "step": 1682
    },
    {
      "epoch": 1.7494802494802495,
      "grad_norm": 4.8014326095581055,
      "learning_rate": 7.2256872256872265e-06,
      "loss": 0.1296,
      "step": 1683
    },
    {
      "epoch": 1.7505197505197505,
      "grad_norm": 6.449215888977051,
      "learning_rate": 7.223377223377224e-06,
      "loss": 0.2195,
      "step": 1684
    },
    {
      "epoch": 1.7515592515592515,
      "grad_norm": 8.582338333129883,
      "learning_rate": 7.221067221067222e-06,
      "loss": 0.261,
      "step": 1685
    },
    {
      "epoch": 1.7525987525987525,
      "grad_norm": 1.6855406761169434,
      "learning_rate": 7.218757218757219e-06,
      "loss": 0.0462,
      "step": 1686
    },
    {
      "epoch": 1.7536382536382535,
      "grad_norm": 0.14350074529647827,
      "learning_rate": 7.216447216447217e-06,
      "loss": 0.002,
      "step": 1687
    },
    {
      "epoch": 1.7546777546777546,
      "grad_norm": 7.426890850067139,
      "learning_rate": 7.214137214137215e-06,
      "loss": 0.2078,
      "step": 1688
    },
    {
      "epoch": 1.7557172557172556,
      "grad_norm": 7.843624591827393,
      "learning_rate": 7.211827211827212e-06,
      "loss": 0.4337,
      "step": 1689
    },
    {
      "epoch": 1.7567567567567568,
      "grad_norm": 2.7187883853912354,
      "learning_rate": 7.20951720951721e-06,
      "loss": 0.032,
      "step": 1690
    },
    {
      "epoch": 1.7577962577962578,
      "grad_norm": 9.4097261428833,
      "learning_rate": 7.207207207207208e-06,
      "loss": 0.5517,
      "step": 1691
    },
    {
      "epoch": 1.7588357588357588,
      "grad_norm": 0.20567497611045837,
      "learning_rate": 7.204897204897205e-06,
      "loss": 0.0046,
      "step": 1692
    },
    {
      "epoch": 1.7598752598752598,
      "grad_norm": 3.7219395637512207,
      "learning_rate": 7.202587202587203e-06,
      "loss": 0.0952,
      "step": 1693
    },
    {
      "epoch": 1.760914760914761,
      "grad_norm": 0.9259941577911377,
      "learning_rate": 7.200277200277201e-06,
      "loss": 0.0224,
      "step": 1694
    },
    {
      "epoch": 1.761954261954262,
      "grad_norm": 0.19360046088695526,
      "learning_rate": 7.197967197967198e-06,
      "loss": 0.0032,
      "step": 1695
    },
    {
      "epoch": 1.762993762993763,
      "grad_norm": 9.575984954833984,
      "learning_rate": 7.195657195657196e-06,
      "loss": 0.9476,
      "step": 1696
    },
    {
      "epoch": 1.764033264033264,
      "grad_norm": 0.27635741233825684,
      "learning_rate": 7.193347193347194e-06,
      "loss": 0.0053,
      "step": 1697
    },
    {
      "epoch": 1.7650727650727651,
      "grad_norm": 5.570565700531006,
      "learning_rate": 7.191037191037191e-06,
      "loss": 0.1848,
      "step": 1698
    },
    {
      "epoch": 1.7661122661122661,
      "grad_norm": 8.489222526550293,
      "learning_rate": 7.188727188727189e-06,
      "loss": 0.327,
      "step": 1699
    },
    {
      "epoch": 1.7671517671517671,
      "grad_norm": 0.45986512303352356,
      "learning_rate": 7.186417186417187e-06,
      "loss": 0.0076,
      "step": 1700
    },
    {
      "epoch": 1.7681912681912682,
      "grad_norm": 1.1088552474975586,
      "learning_rate": 7.1841071841071854e-06,
      "loss": 0.0475,
      "step": 1701
    },
    {
      "epoch": 1.7692307692307692,
      "grad_norm": 13.177712440490723,
      "learning_rate": 7.181797181797182e-06,
      "loss": 0.203,
      "step": 1702
    },
    {
      "epoch": 1.7702702702702702,
      "grad_norm": 7.197229862213135,
      "learning_rate": 7.17948717948718e-06,
      "loss": 0.4223,
      "step": 1703
    },
    {
      "epoch": 1.7713097713097712,
      "grad_norm": 3.42079496383667,
      "learning_rate": 7.177177177177178e-06,
      "loss": 0.0728,
      "step": 1704
    },
    {
      "epoch": 1.7723492723492722,
      "grad_norm": 0.43769392371177673,
      "learning_rate": 7.174867174867175e-06,
      "loss": 0.0143,
      "step": 1705
    },
    {
      "epoch": 1.7733887733887734,
      "grad_norm": 0.10485343635082245,
      "learning_rate": 7.172557172557173e-06,
      "loss": 0.0023,
      "step": 1706
    },
    {
      "epoch": 1.7744282744282744,
      "grad_norm": 9.893296241760254,
      "learning_rate": 7.170247170247171e-06,
      "loss": 0.8923,
      "step": 1707
    },
    {
      "epoch": 1.7754677754677755,
      "grad_norm": 4.010193347930908,
      "learning_rate": 7.167937167937169e-06,
      "loss": 0.1206,
      "step": 1708
    },
    {
      "epoch": 1.7765072765072765,
      "grad_norm": 0.3834474980831146,
      "learning_rate": 7.165627165627166e-06,
      "loss": 0.0048,
      "step": 1709
    },
    {
      "epoch": 1.7775467775467777,
      "grad_norm": 3.8938255310058594,
      "learning_rate": 7.163317163317164e-06,
      "loss": 0.1693,
      "step": 1710
    },
    {
      "epoch": 1.7785862785862787,
      "grad_norm": 11.818151473999023,
      "learning_rate": 7.161007161007162e-06,
      "loss": 0.4191,
      "step": 1711
    },
    {
      "epoch": 1.7796257796257797,
      "grad_norm": 2.825345039367676,
      "learning_rate": 7.158697158697159e-06,
      "loss": 0.0724,
      "step": 1712
    },
    {
      "epoch": 1.7806652806652807,
      "grad_norm": 0.23397642374038696,
      "learning_rate": 7.156387156387157e-06,
      "loss": 0.0085,
      "step": 1713
    },
    {
      "epoch": 1.7817047817047817,
      "grad_norm": 2.670799970626831,
      "learning_rate": 7.154077154077155e-06,
      "loss": 0.0735,
      "step": 1714
    },
    {
      "epoch": 1.7827442827442828,
      "grad_norm": 6.973132133483887,
      "learning_rate": 7.1517671517671525e-06,
      "loss": 0.3394,
      "step": 1715
    },
    {
      "epoch": 1.7837837837837838,
      "grad_norm": 5.924585342407227,
      "learning_rate": 7.14945714945715e-06,
      "loss": 0.3206,
      "step": 1716
    },
    {
      "epoch": 1.7848232848232848,
      "grad_norm": 7.079158306121826,
      "learning_rate": 7.147147147147148e-06,
      "loss": 0.9937,
      "step": 1717
    },
    {
      "epoch": 1.7858627858627858,
      "grad_norm": 0.8085556626319885,
      "learning_rate": 7.144837144837145e-06,
      "loss": 0.0233,
      "step": 1718
    },
    {
      "epoch": 1.7869022869022868,
      "grad_norm": 0.26186278462409973,
      "learning_rate": 7.1425271425271435e-06,
      "loss": 0.0043,
      "step": 1719
    },
    {
      "epoch": 1.7879417879417878,
      "grad_norm": 3.6057870388031006,
      "learning_rate": 7.140217140217141e-06,
      "loss": 0.1303,
      "step": 1720
    },
    {
      "epoch": 1.7889812889812888,
      "grad_norm": 7.762540340423584,
      "learning_rate": 7.137907137907138e-06,
      "loss": 0.6902,
      "step": 1721
    },
    {
      "epoch": 1.79002079002079,
      "grad_norm": 14.205892562866211,
      "learning_rate": 7.1355971355971364e-06,
      "loss": 2.8257,
      "step": 1722
    },
    {
      "epoch": 1.791060291060291,
      "grad_norm": 11.998815536499023,
      "learning_rate": 7.133287133287134e-06,
      "loss": 1.8192,
      "step": 1723
    },
    {
      "epoch": 1.792099792099792,
      "grad_norm": 4.715017318725586,
      "learning_rate": 7.130977130977131e-06,
      "loss": 0.1902,
      "step": 1724
    },
    {
      "epoch": 1.793139293139293,
      "grad_norm": 6.3350396156311035,
      "learning_rate": 7.128667128667129e-06,
      "loss": 0.1873,
      "step": 1725
    },
    {
      "epoch": 1.7941787941787943,
      "grad_norm": 9.118733406066895,
      "learning_rate": 7.1263571263571275e-06,
      "loss": 0.56,
      "step": 1726
    },
    {
      "epoch": 1.7952182952182953,
      "grad_norm": 5.499832630157471,
      "learning_rate": 7.124047124047124e-06,
      "loss": 0.3405,
      "step": 1727
    },
    {
      "epoch": 1.7962577962577964,
      "grad_norm": 1.0076048374176025,
      "learning_rate": 7.121737121737122e-06,
      "loss": 0.0391,
      "step": 1728
    },
    {
      "epoch": 1.7972972972972974,
      "grad_norm": 9.98312759399414,
      "learning_rate": 7.11942711942712e-06,
      "loss": 1.3863,
      "step": 1729
    },
    {
      "epoch": 1.7983367983367984,
      "grad_norm": 6.568471908569336,
      "learning_rate": 7.117117117117117e-06,
      "loss": 0.356,
      "step": 1730
    },
    {
      "epoch": 1.7993762993762994,
      "grad_norm": 8.882364273071289,
      "learning_rate": 7.114807114807115e-06,
      "loss": 0.5891,
      "step": 1731
    },
    {
      "epoch": 1.8004158004158004,
      "grad_norm": 0.11736893653869629,
      "learning_rate": 7.112497112497113e-06,
      "loss": 0.003,
      "step": 1732
    },
    {
      "epoch": 1.8014553014553014,
      "grad_norm": 11.34277057647705,
      "learning_rate": 7.1101871101871114e-06,
      "loss": 0.0025,
      "step": 1733
    },
    {
      "epoch": 1.8024948024948024,
      "grad_norm": 0.47979533672332764,
      "learning_rate": 7.107877107877108e-06,
      "loss": 0.0094,
      "step": 1734
    },
    {
      "epoch": 1.8035343035343034,
      "grad_norm": 7.334555149078369,
      "learning_rate": 7.105567105567106e-06,
      "loss": 0.3448,
      "step": 1735
    },
    {
      "epoch": 1.8045738045738045,
      "grad_norm": 7.494808197021484,
      "learning_rate": 7.103257103257104e-06,
      "loss": 0.8725,
      "step": 1736
    },
    {
      "epoch": 1.8056133056133055,
      "grad_norm": 4.283798694610596,
      "learning_rate": 7.100947100947102e-06,
      "loss": 0.1439,
      "step": 1737
    },
    {
      "epoch": 1.8066528066528067,
      "grad_norm": 1.145999789237976,
      "learning_rate": 7.098637098637099e-06,
      "loss": 0.025,
      "step": 1738
    },
    {
      "epoch": 1.8076923076923077,
      "grad_norm": 0.05247219651937485,
      "learning_rate": 7.096327096327097e-06,
      "loss": 0.001,
      "step": 1739
    },
    {
      "epoch": 1.8087318087318087,
      "grad_norm": 2.770738124847412,
      "learning_rate": 7.0940170940170945e-06,
      "loss": 0.1025,
      "step": 1740
    },
    {
      "epoch": 1.8097713097713097,
      "grad_norm": 1.1891086101531982,
      "learning_rate": 7.091707091707092e-06,
      "loss": 0.0212,
      "step": 1741
    },
    {
      "epoch": 1.810810810810811,
      "grad_norm": 1.8017946481704712,
      "learning_rate": 7.08939708939709e-06,
      "loss": 0.0425,
      "step": 1742
    },
    {
      "epoch": 1.811850311850312,
      "grad_norm": 3.020580530166626,
      "learning_rate": 7.087087087087087e-06,
      "loss": 0.0616,
      "step": 1743
    },
    {
      "epoch": 1.812889812889813,
      "grad_norm": 3.8289687633514404,
      "learning_rate": 7.084777084777086e-06,
      "loss": 0.155,
      "step": 1744
    },
    {
      "epoch": 1.813929313929314,
      "grad_norm": 0.27335742115974426,
      "learning_rate": 7.082467082467083e-06,
      "loss": 0.0069,
      "step": 1745
    },
    {
      "epoch": 1.814968814968815,
      "grad_norm": 9.197246551513672,
      "learning_rate": 7.08015708015708e-06,
      "loss": 0.1974,
      "step": 1746
    },
    {
      "epoch": 1.816008316008316,
      "grad_norm": 4.551156997680664,
      "learning_rate": 7.0778470778470785e-06,
      "loss": 0.0991,
      "step": 1747
    },
    {
      "epoch": 1.817047817047817,
      "grad_norm": 1.667893409729004,
      "learning_rate": 7.075537075537076e-06,
      "loss": 0.0315,
      "step": 1748
    },
    {
      "epoch": 1.818087318087318,
      "grad_norm": 15.546823501586914,
      "learning_rate": 7.073227073227073e-06,
      "loss": 1.4498,
      "step": 1749
    },
    {
      "epoch": 1.819126819126819,
      "grad_norm": 6.499513626098633,
      "learning_rate": 7.070917070917071e-06,
      "loss": 0.1394,
      "step": 1750
    },
    {
      "epoch": 1.82016632016632,
      "grad_norm": 3.992257833480835,
      "learning_rate": 7.0686070686070696e-06,
      "loss": 0.1193,
      "step": 1751
    },
    {
      "epoch": 1.821205821205821,
      "grad_norm": 2.7317323684692383,
      "learning_rate": 7.066297066297066e-06,
      "loss": 0.0496,
      "step": 1752
    },
    {
      "epoch": 1.822245322245322,
      "grad_norm": 0.6675400137901306,
      "learning_rate": 7.063987063987064e-06,
      "loss": 0.0152,
      "step": 1753
    },
    {
      "epoch": 1.8232848232848233,
      "grad_norm": 0.47929680347442627,
      "learning_rate": 7.0616770616770624e-06,
      "loss": 0.0178,
      "step": 1754
    },
    {
      "epoch": 1.8243243243243243,
      "grad_norm": 5.309060573577881,
      "learning_rate": 7.059367059367061e-06,
      "loss": 0.1906,
      "step": 1755
    },
    {
      "epoch": 1.8253638253638254,
      "grad_norm": 0.5200766324996948,
      "learning_rate": 7.057057057057057e-06,
      "loss": 0.0098,
      "step": 1756
    },
    {
      "epoch": 1.8264033264033264,
      "grad_norm": 3.534630060195923,
      "learning_rate": 7.054747054747055e-06,
      "loss": 0.1054,
      "step": 1757
    },
    {
      "epoch": 1.8274428274428276,
      "grad_norm": 8.29762077331543,
      "learning_rate": 7.0524370524370535e-06,
      "loss": 0.3073,
      "step": 1758
    },
    {
      "epoch": 1.8284823284823286,
      "grad_norm": 0.8621770143508911,
      "learning_rate": 7.05012705012705e-06,
      "loss": 0.0204,
      "step": 1759
    },
    {
      "epoch": 1.8295218295218296,
      "grad_norm": 0.23380033671855927,
      "learning_rate": 7.047817047817048e-06,
      "loss": 0.0035,
      "step": 1760
    },
    {
      "epoch": 1.8305613305613306,
      "grad_norm": 5.623581409454346,
      "learning_rate": 7.045507045507046e-06,
      "loss": 0.2227,
      "step": 1761
    },
    {
      "epoch": 1.8316008316008316,
      "grad_norm": 2.2980833053588867,
      "learning_rate": 7.0431970431970446e-06,
      "loss": 0.0643,
      "step": 1762
    },
    {
      "epoch": 1.8326403326403327,
      "grad_norm": 12.151287078857422,
      "learning_rate": 7.040887040887041e-06,
      "loss": 0.2851,
      "step": 1763
    },
    {
      "epoch": 1.8336798336798337,
      "grad_norm": 0.5834892392158508,
      "learning_rate": 7.038577038577039e-06,
      "loss": 0.0108,
      "step": 1764
    },
    {
      "epoch": 1.8347193347193347,
      "grad_norm": 0.06696605682373047,
      "learning_rate": 7.0362670362670374e-06,
      "loss": 0.0014,
      "step": 1765
    },
    {
      "epoch": 1.8357588357588357,
      "grad_norm": 4.016539096832275,
      "learning_rate": 7.033957033957034e-06,
      "loss": 0.1166,
      "step": 1766
    },
    {
      "epoch": 1.8367983367983367,
      "grad_norm": 1.0611064434051514,
      "learning_rate": 7.031647031647032e-06,
      "loss": 0.041,
      "step": 1767
    },
    {
      "epoch": 1.8378378378378377,
      "grad_norm": 0.7719981074333191,
      "learning_rate": 7.02933702933703e-06,
      "loss": 0.0163,
      "step": 1768
    },
    {
      "epoch": 1.8388773388773387,
      "grad_norm": 3.026901960372925,
      "learning_rate": 7.027027027027028e-06,
      "loss": 0.0903,
      "step": 1769
    },
    {
      "epoch": 1.83991683991684,
      "grad_norm": 3.5234310626983643,
      "learning_rate": 7.024717024717025e-06,
      "loss": 0.111,
      "step": 1770
    },
    {
      "epoch": 1.840956340956341,
      "grad_norm": 0.18749161064624786,
      "learning_rate": 7.022407022407023e-06,
      "loss": 0.003,
      "step": 1771
    },
    {
      "epoch": 1.841995841995842,
      "grad_norm": 1.624272108078003,
      "learning_rate": 7.0200970200970205e-06,
      "loss": 0.0276,
      "step": 1772
    },
    {
      "epoch": 1.843035343035343,
      "grad_norm": 0.39288201928138733,
      "learning_rate": 7.017787017787019e-06,
      "loss": 0.0139,
      "step": 1773
    },
    {
      "epoch": 1.8440748440748442,
      "grad_norm": 8.802492141723633,
      "learning_rate": 7.015477015477016e-06,
      "loss": 0.0945,
      "step": 1774
    },
    {
      "epoch": 1.8451143451143452,
      "grad_norm": 10.572566986083984,
      "learning_rate": 7.0131670131670134e-06,
      "loss": 0.399,
      "step": 1775
    },
    {
      "epoch": 1.8461538461538463,
      "grad_norm": 6.584804058074951,
      "learning_rate": 7.010857010857012e-06,
      "loss": 0.3771,
      "step": 1776
    },
    {
      "epoch": 1.8471933471933473,
      "grad_norm": 2.911133289337158,
      "learning_rate": 7.008547008547009e-06,
      "loss": 0.0747,
      "step": 1777
    },
    {
      "epoch": 1.8482328482328483,
      "grad_norm": 0.4883025288581848,
      "learning_rate": 7.006237006237006e-06,
      "loss": 0.0088,
      "step": 1778
    },
    {
      "epoch": 1.8492723492723493,
      "grad_norm": 10.067765235900879,
      "learning_rate": 7.0039270039270045e-06,
      "loss": 0.6624,
      "step": 1779
    },
    {
      "epoch": 1.8503118503118503,
      "grad_norm": 1.677093505859375,
      "learning_rate": 7.001617001617003e-06,
      "loss": 0.0389,
      "step": 1780
    },
    {
      "epoch": 1.8513513513513513,
      "grad_norm": 6.398299217224121,
      "learning_rate": 6.999306999306999e-06,
      "loss": 0.365,
      "step": 1781
    },
    {
      "epoch": 1.8523908523908523,
      "grad_norm": 0.02928902953863144,
      "learning_rate": 6.996996996996997e-06,
      "loss": 0.001,
      "step": 1782
    },
    {
      "epoch": 1.8534303534303533,
      "grad_norm": 13.064369201660156,
      "learning_rate": 6.9946869946869956e-06,
      "loss": 0.9939,
      "step": 1783
    },
    {
      "epoch": 1.8544698544698544,
      "grad_norm": 0.872677743434906,
      "learning_rate": 6.992376992376992e-06,
      "loss": 0.0126,
      "step": 1784
    },
    {
      "epoch": 1.8555093555093554,
      "grad_norm": 4.7186198234558105,
      "learning_rate": 6.99006699006699e-06,
      "loss": 0.2806,
      "step": 1785
    },
    {
      "epoch": 1.8565488565488566,
      "grad_norm": 2.009587287902832,
      "learning_rate": 6.9877569877569884e-06,
      "loss": 0.1332,
      "step": 1786
    },
    {
      "epoch": 1.8575883575883576,
      "grad_norm": 10.572334289550781,
      "learning_rate": 6.985446985446987e-06,
      "loss": 0.6191,
      "step": 1787
    },
    {
      "epoch": 1.8586278586278586,
      "grad_norm": 2.9769487380981445,
      "learning_rate": 6.983136983136983e-06,
      "loss": 0.0616,
      "step": 1788
    },
    {
      "epoch": 1.8596673596673596,
      "grad_norm": 6.331596374511719,
      "learning_rate": 6.980826980826981e-06,
      "loss": 0.2601,
      "step": 1789
    },
    {
      "epoch": 1.8607068607068609,
      "grad_norm": 0.20288600027561188,
      "learning_rate": 6.9785169785169795e-06,
      "loss": 0.0061,
      "step": 1790
    },
    {
      "epoch": 1.8617463617463619,
      "grad_norm": 8.505599021911621,
      "learning_rate": 6.976206976206978e-06,
      "loss": 0.0917,
      "step": 1791
    },
    {
      "epoch": 1.862785862785863,
      "grad_norm": 1.9080631732940674,
      "learning_rate": 6.973896973896974e-06,
      "loss": 0.0389,
      "step": 1792
    },
    {
      "epoch": 1.863825363825364,
      "grad_norm": 1.8575496673583984,
      "learning_rate": 6.971586971586972e-06,
      "loss": 0.039,
      "step": 1793
    },
    {
      "epoch": 1.864864864864865,
      "grad_norm": 2.9888617992401123,
      "learning_rate": 6.969276969276971e-06,
      "loss": 0.0925,
      "step": 1794
    },
    {
      "epoch": 1.865904365904366,
      "grad_norm": 13.349923133850098,
      "learning_rate": 6.966966966966967e-06,
      "loss": 0.9717,
      "step": 1795
    },
    {
      "epoch": 1.866943866943867,
      "grad_norm": 6.0613861083984375,
      "learning_rate": 6.964656964656965e-06,
      "loss": 0.2833,
      "step": 1796
    },
    {
      "epoch": 1.867983367983368,
      "grad_norm": 2.117635726928711,
      "learning_rate": 6.9623469623469635e-06,
      "loss": 0.0424,
      "step": 1797
    },
    {
      "epoch": 1.869022869022869,
      "grad_norm": 2.256873607635498,
      "learning_rate": 6.960036960036961e-06,
      "loss": 0.0965,
      "step": 1798
    },
    {
      "epoch": 1.87006237006237,
      "grad_norm": 0.33173665404319763,
      "learning_rate": 6.957726957726958e-06,
      "loss": 0.0106,
      "step": 1799
    },
    {
      "epoch": 1.871101871101871,
      "grad_norm": 4.786936283111572,
      "learning_rate": 6.955416955416956e-06,
      "loss": 0.1814,
      "step": 1800
    },
    {
      "epoch": 1.872141372141372,
      "grad_norm": 0.11546695232391357,
      "learning_rate": 6.953106953106954e-06,
      "loss": 0.0017,
      "step": 1801
    },
    {
      "epoch": 1.8731808731808732,
      "grad_norm": 5.822028160095215,
      "learning_rate": 6.950796950796951e-06,
      "loss": 0.132,
      "step": 1802
    },
    {
      "epoch": 1.8742203742203742,
      "grad_norm": 29.958877563476562,
      "learning_rate": 6.948486948486949e-06,
      "loss": 0.6952,
      "step": 1803
    },
    {
      "epoch": 1.8752598752598753,
      "grad_norm": 17.843074798583984,
      "learning_rate": 6.9461769461769466e-06,
      "loss": 1.4509,
      "step": 1804
    },
    {
      "epoch": 1.8762993762993763,
      "grad_norm": 7.777376174926758,
      "learning_rate": 6.943866943866945e-06,
      "loss": 0.2881,
      "step": 1805
    },
    {
      "epoch": 1.8773388773388775,
      "grad_norm": 3.4465243816375732,
      "learning_rate": 6.941556941556942e-06,
      "loss": 0.0959,
      "step": 1806
    },
    {
      "epoch": 1.8783783783783785,
      "grad_norm": 0.019290627911686897,
      "learning_rate": 6.9392469392469394e-06,
      "loss": 0.0005,
      "step": 1807
    },
    {
      "epoch": 1.8794178794178795,
      "grad_norm": 0.029469970613718033,
      "learning_rate": 6.936936936936938e-06,
      "loss": 0.0007,
      "step": 1808
    },
    {
      "epoch": 1.8804573804573805,
      "grad_norm": 7.679044723510742,
      "learning_rate": 6.934626934626936e-06,
      "loss": 0.4403,
      "step": 1809
    },
    {
      "epoch": 1.8814968814968815,
      "grad_norm": 0.09692568331956863,
      "learning_rate": 6.932316932316932e-06,
      "loss": 0.0024,
      "step": 1810
    },
    {
      "epoch": 1.8825363825363826,
      "grad_norm": 5.558783054351807,
      "learning_rate": 6.9300069300069305e-06,
      "loss": 0.1236,
      "step": 1811
    },
    {
      "epoch": 1.8835758835758836,
      "grad_norm": 9.43947982788086,
      "learning_rate": 6.927696927696929e-06,
      "loss": 0.4962,
      "step": 1812
    },
    {
      "epoch": 1.8846153846153846,
      "grad_norm": 1.136598825454712,
      "learning_rate": 6.925386925386925e-06,
      "loss": 0.0386,
      "step": 1813
    },
    {
      "epoch": 1.8856548856548856,
      "grad_norm": 3.2195546627044678,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.0834,
      "step": 1814
    },
    {
      "epoch": 1.8866943866943866,
      "grad_norm": 6.025984287261963,
      "learning_rate": 6.9207669207669216e-06,
      "loss": 0.3229,
      "step": 1815
    },
    {
      "epoch": 1.8877338877338876,
      "grad_norm": 1.9402605295181274,
      "learning_rate": 6.91845691845692e-06,
      "loss": 0.0161,
      "step": 1816
    },
    {
      "epoch": 1.8887733887733886,
      "grad_norm": 1.3776854276657104,
      "learning_rate": 6.916146916146916e-06,
      "loss": 0.0468,
      "step": 1817
    },
    {
      "epoch": 1.8898128898128899,
      "grad_norm": 0.0545632541179657,
      "learning_rate": 6.9138369138369145e-06,
      "loss": 0.0011,
      "step": 1818
    },
    {
      "epoch": 1.8908523908523909,
      "grad_norm": 2.274041175842285,
      "learning_rate": 6.911526911526913e-06,
      "loss": 0.0631,
      "step": 1819
    },
    {
      "epoch": 1.8918918918918919,
      "grad_norm": 2.384172201156616,
      "learning_rate": 6.909216909216909e-06,
      "loss": 0.0757,
      "step": 1820
    },
    {
      "epoch": 1.892931392931393,
      "grad_norm": 8.121810913085938,
      "learning_rate": 6.906906906906907e-06,
      "loss": 0.2988,
      "step": 1821
    },
    {
      "epoch": 1.8939708939708941,
      "grad_norm": 7.533204078674316,
      "learning_rate": 6.9045969045969055e-06,
      "loss": 0.7463,
      "step": 1822
    },
    {
      "epoch": 1.8950103950103951,
      "grad_norm": 7.793642520904541,
      "learning_rate": 6.902286902286903e-06,
      "loss": 0.3933,
      "step": 1823
    },
    {
      "epoch": 1.8960498960498962,
      "grad_norm": 6.824788570404053,
      "learning_rate": 6.8999768999769e-06,
      "loss": 0.1922,
      "step": 1824
    },
    {
      "epoch": 1.8970893970893972,
      "grad_norm": 4.8206915855407715,
      "learning_rate": 6.897666897666898e-06,
      "loss": 0.3425,
      "step": 1825
    },
    {
      "epoch": 1.8981288981288982,
      "grad_norm": 12.999703407287598,
      "learning_rate": 6.895356895356896e-06,
      "loss": 0.4453,
      "step": 1826
    },
    {
      "epoch": 1.8991683991683992,
      "grad_norm": 7.11517333984375,
      "learning_rate": 6.893046893046894e-06,
      "loss": 0.343,
      "step": 1827
    },
    {
      "epoch": 1.9002079002079002,
      "grad_norm": 0.7533423900604248,
      "learning_rate": 6.890736890736891e-06,
      "loss": 0.0221,
      "step": 1828
    },
    {
      "epoch": 1.9012474012474012,
      "grad_norm": 8.051311492919922,
      "learning_rate": 6.888426888426889e-06,
      "loss": 0.6874,
      "step": 1829
    },
    {
      "epoch": 1.9022869022869022,
      "grad_norm": 1.4836094379425049,
      "learning_rate": 6.886116886116887e-06,
      "loss": 0.0364,
      "step": 1830
    },
    {
      "epoch": 1.9033264033264032,
      "grad_norm": 9.048811912536621,
      "learning_rate": 6.883806883806884e-06,
      "loss": 0.4148,
      "step": 1831
    },
    {
      "epoch": 1.9043659043659042,
      "grad_norm": 0.09245787560939789,
      "learning_rate": 6.8814968814968815e-06,
      "loss": 0.0034,
      "step": 1832
    },
    {
      "epoch": 1.9054054054054053,
      "grad_norm": 5.148186683654785,
      "learning_rate": 6.87918687918688e-06,
      "loss": 0.1343,
      "step": 1833
    },
    {
      "epoch": 1.9064449064449065,
      "grad_norm": 0.022289017215371132,
      "learning_rate": 6.876876876876878e-06,
      "loss": 0.0005,
      "step": 1834
    },
    {
      "epoch": 1.9074844074844075,
      "grad_norm": 15.062718391418457,
      "learning_rate": 6.874566874566874e-06,
      "loss": 0.6481,
      "step": 1835
    },
    {
      "epoch": 1.9085239085239085,
      "grad_norm": 0.6894482374191284,
      "learning_rate": 6.8722568722568726e-06,
      "loss": 0.0068,
      "step": 1836
    },
    {
      "epoch": 1.9095634095634095,
      "grad_norm": 7.909749984741211,
      "learning_rate": 6.869946869946871e-06,
      "loss": 0.5693,
      "step": 1837
    },
    {
      "epoch": 1.9106029106029108,
      "grad_norm": 5.598539352416992,
      "learning_rate": 6.867636867636867e-06,
      "loss": 0.2707,
      "step": 1838
    },
    {
      "epoch": 1.9116424116424118,
      "grad_norm": 5.760708332061768,
      "learning_rate": 6.8653268653268654e-06,
      "loss": 0.1846,
      "step": 1839
    },
    {
      "epoch": 1.9126819126819128,
      "grad_norm": 3.930959463119507,
      "learning_rate": 6.863016863016864e-06,
      "loss": 0.0626,
      "step": 1840
    },
    {
      "epoch": 1.9137214137214138,
      "grad_norm": 0.062302954494953156,
      "learning_rate": 6.860706860706862e-06,
      "loss": 0.0013,
      "step": 1841
    },
    {
      "epoch": 1.9147609147609148,
      "grad_norm": 2.7454307079315186,
      "learning_rate": 6.858396858396858e-06,
      "loss": 0.0692,
      "step": 1842
    },
    {
      "epoch": 1.9158004158004158,
      "grad_norm": 3.4849820137023926,
      "learning_rate": 6.8560868560868565e-06,
      "loss": 0.1363,
      "step": 1843
    },
    {
      "epoch": 1.9168399168399168,
      "grad_norm": 8.270879745483398,
      "learning_rate": 6.853776853776855e-06,
      "loss": 0.4004,
      "step": 1844
    },
    {
      "epoch": 1.9178794178794178,
      "grad_norm": 1.630702018737793,
      "learning_rate": 6.851466851466853e-06,
      "loss": 0.046,
      "step": 1845
    },
    {
      "epoch": 1.9189189189189189,
      "grad_norm": 0.4970608353614807,
      "learning_rate": 6.849156849156849e-06,
      "loss": 0.009,
      "step": 1846
    },
    {
      "epoch": 1.9199584199584199,
      "grad_norm": 0.20573386549949646,
      "learning_rate": 6.846846846846848e-06,
      "loss": 0.0041,
      "step": 1847
    },
    {
      "epoch": 1.9209979209979209,
      "grad_norm": 1.271240472793579,
      "learning_rate": 6.844536844536846e-06,
      "loss": 0.0353,
      "step": 1848
    },
    {
      "epoch": 1.922037422037422,
      "grad_norm": 0.43870508670806885,
      "learning_rate": 6.842226842226842e-06,
      "loss": 0.0148,
      "step": 1849
    },
    {
      "epoch": 1.9230769230769231,
      "grad_norm": 7.131784439086914,
      "learning_rate": 6.8399168399168405e-06,
      "loss": 0.3034,
      "step": 1850
    },
    {
      "epoch": 1.9241164241164241,
      "grad_norm": 5.150654315948486,
      "learning_rate": 6.837606837606839e-06,
      "loss": 0.1394,
      "step": 1851
    },
    {
      "epoch": 1.9251559251559252,
      "grad_norm": 13.46895694732666,
      "learning_rate": 6.835296835296836e-06,
      "loss": 1.7075,
      "step": 1852
    },
    {
      "epoch": 1.9261954261954262,
      "grad_norm": 0.38688424229621887,
      "learning_rate": 6.832986832986833e-06,
      "loss": 0.0091,
      "step": 1853
    },
    {
      "epoch": 1.9272349272349274,
      "grad_norm": 6.538180828094482,
      "learning_rate": 6.8306768306768315e-06,
      "loss": 0.7274,
      "step": 1854
    },
    {
      "epoch": 1.9282744282744284,
      "grad_norm": 5.841011047363281,
      "learning_rate": 6.828366828366829e-06,
      "loss": 0.1061,
      "step": 1855
    },
    {
      "epoch": 1.9293139293139294,
      "grad_norm": 10.935325622558594,
      "learning_rate": 6.826056826056826e-06,
      "loss": 1.1941,
      "step": 1856
    },
    {
      "epoch": 1.9303534303534304,
      "grad_norm": 10.663113594055176,
      "learning_rate": 6.823746823746824e-06,
      "loss": 0.792,
      "step": 1857
    },
    {
      "epoch": 1.9313929313929314,
      "grad_norm": 3.2219295501708984,
      "learning_rate": 6.821436821436822e-06,
      "loss": 0.1102,
      "step": 1858
    },
    {
      "epoch": 1.9324324324324325,
      "grad_norm": 0.5760859847068787,
      "learning_rate": 6.81912681912682e-06,
      "loss": 0.0083,
      "step": 1859
    },
    {
      "epoch": 1.9334719334719335,
      "grad_norm": 4.954216003417969,
      "learning_rate": 6.816816816816817e-06,
      "loss": 0.1843,
      "step": 1860
    },
    {
      "epoch": 1.9345114345114345,
      "grad_norm": 10.934431076049805,
      "learning_rate": 6.814506814506815e-06,
      "loss": 0.5659,
      "step": 1861
    },
    {
      "epoch": 1.9355509355509355,
      "grad_norm": 6.139547824859619,
      "learning_rate": 6.812196812196813e-06,
      "loss": 0.3543,
      "step": 1862
    },
    {
      "epoch": 1.9365904365904365,
      "grad_norm": 4.316939830780029,
      "learning_rate": 6.809886809886811e-06,
      "loss": 0.1155,
      "step": 1863
    },
    {
      "epoch": 1.9376299376299375,
      "grad_norm": 9.48974895477295,
      "learning_rate": 6.8075768075768075e-06,
      "loss": 0.64,
      "step": 1864
    },
    {
      "epoch": 1.9386694386694385,
      "grad_norm": 1.2552036046981812,
      "learning_rate": 6.805266805266806e-06,
      "loss": 0.0173,
      "step": 1865
    },
    {
      "epoch": 1.9397089397089398,
      "grad_norm": 0.9558032155036926,
      "learning_rate": 6.802956802956804e-06,
      "loss": 0.0376,
      "step": 1866
    },
    {
      "epoch": 1.9407484407484408,
      "grad_norm": 1.8135472536087036,
      "learning_rate": 6.8006468006468e-06,
      "loss": 0.0469,
      "step": 1867
    },
    {
      "epoch": 1.9417879417879418,
      "grad_norm": 7.1318511962890625,
      "learning_rate": 6.7983367983367986e-06,
      "loss": 0.7816,
      "step": 1868
    },
    {
      "epoch": 1.9428274428274428,
      "grad_norm": 0.13948020339012146,
      "learning_rate": 6.796026796026797e-06,
      "loss": 0.0017,
      "step": 1869
    },
    {
      "epoch": 1.943866943866944,
      "grad_norm": 0.0776764452457428,
      "learning_rate": 6.793716793716795e-06,
      "loss": 0.0011,
      "step": 1870
    },
    {
      "epoch": 1.944906444906445,
      "grad_norm": 0.2135138213634491,
      "learning_rate": 6.7914067914067915e-06,
      "loss": 0.0061,
      "step": 1871
    },
    {
      "epoch": 1.945945945945946,
      "grad_norm": 0.08584604412317276,
      "learning_rate": 6.78909678909679e-06,
      "loss": 0.0027,
      "step": 1872
    },
    {
      "epoch": 1.946985446985447,
      "grad_norm": 4.477102279663086,
      "learning_rate": 6.786786786786788e-06,
      "loss": 0.1093,
      "step": 1873
    },
    {
      "epoch": 1.948024948024948,
      "grad_norm": 0.3456413745880127,
      "learning_rate": 6.784476784476784e-06,
      "loss": 0.0076,
      "step": 1874
    },
    {
      "epoch": 1.949064449064449,
      "grad_norm": 4.4420013427734375,
      "learning_rate": 6.7821667821667825e-06,
      "loss": 0.2816,
      "step": 1875
    },
    {
      "epoch": 1.95010395010395,
      "grad_norm": 9.357203483581543,
      "learning_rate": 6.779856779856781e-06,
      "loss": 0.3661,
      "step": 1876
    },
    {
      "epoch": 1.9511434511434511,
      "grad_norm": 11.721562385559082,
      "learning_rate": 6.777546777546778e-06,
      "loss": 0.8204,
      "step": 1877
    },
    {
      "epoch": 1.9521829521829521,
      "grad_norm": 0.9358542561531067,
      "learning_rate": 6.775236775236775e-06,
      "loss": 0.0355,
      "step": 1878
    },
    {
      "epoch": 1.9532224532224531,
      "grad_norm": 10.567963600158691,
      "learning_rate": 6.772926772926774e-06,
      "loss": 0.4664,
      "step": 1879
    },
    {
      "epoch": 1.9542619542619541,
      "grad_norm": 3.096851348876953,
      "learning_rate": 6.770616770616772e-06,
      "loss": 0.0477,
      "step": 1880
    },
    {
      "epoch": 1.9553014553014552,
      "grad_norm": 6.202308177947998,
      "learning_rate": 6.768306768306769e-06,
      "loss": 0.4278,
      "step": 1881
    },
    {
      "epoch": 1.9563409563409564,
      "grad_norm": 1.1024806499481201,
      "learning_rate": 6.7659967659967665e-06,
      "loss": 0.0193,
      "step": 1882
    },
    {
      "epoch": 1.9573804573804574,
      "grad_norm": 0.1680949330329895,
      "learning_rate": 6.763686763686765e-06,
      "loss": 0.0034,
      "step": 1883
    },
    {
      "epoch": 1.9584199584199584,
      "grad_norm": 0.901910126209259,
      "learning_rate": 6.761376761376762e-06,
      "loss": 0.0241,
      "step": 1884
    },
    {
      "epoch": 1.9594594594594594,
      "grad_norm": 2.4601571559906006,
      "learning_rate": 6.759066759066759e-06,
      "loss": 0.0994,
      "step": 1885
    },
    {
      "epoch": 1.9604989604989607,
      "grad_norm": 4.021463871002197,
      "learning_rate": 6.7567567567567575e-06,
      "loss": 0.07,
      "step": 1886
    },
    {
      "epoch": 1.9615384615384617,
      "grad_norm": 0.061176471412181854,
      "learning_rate": 6.754446754446755e-06,
      "loss": 0.0015,
      "step": 1887
    },
    {
      "epoch": 1.9625779625779627,
      "grad_norm": 4.875643730163574,
      "learning_rate": 6.752136752136753e-06,
      "loss": 0.251,
      "step": 1888
    },
    {
      "epoch": 1.9636174636174637,
      "grad_norm": 2.630978584289551,
      "learning_rate": 6.74982674982675e-06,
      "loss": 0.0967,
      "step": 1889
    },
    {
      "epoch": 1.9646569646569647,
      "grad_norm": 1.9153410196304321,
      "learning_rate": 6.747516747516748e-06,
      "loss": 0.054,
      "step": 1890
    },
    {
      "epoch": 1.9656964656964657,
      "grad_norm": 0.2559313178062439,
      "learning_rate": 6.745206745206746e-06,
      "loss": 0.0044,
      "step": 1891
    },
    {
      "epoch": 1.9667359667359667,
      "grad_norm": 6.092691421508789,
      "learning_rate": 6.742896742896743e-06,
      "loss": 0.2965,
      "step": 1892
    },
    {
      "epoch": 1.9677754677754677,
      "grad_norm": 2.28495454788208,
      "learning_rate": 6.740586740586741e-06,
      "loss": 0.0319,
      "step": 1893
    },
    {
      "epoch": 1.9688149688149688,
      "grad_norm": 0.785499095916748,
      "learning_rate": 6.738276738276739e-06,
      "loss": 0.0205,
      "step": 1894
    },
    {
      "epoch": 1.9698544698544698,
      "grad_norm": 7.484012603759766,
      "learning_rate": 6.735966735966737e-06,
      "loss": 0.3765,
      "step": 1895
    },
    {
      "epoch": 1.9708939708939708,
      "grad_norm": 0.0553046315908432,
      "learning_rate": 6.7336567336567335e-06,
      "loss": 0.0015,
      "step": 1896
    },
    {
      "epoch": 1.9719334719334718,
      "grad_norm": 3.8778178691864014,
      "learning_rate": 6.731346731346732e-06,
      "loss": 0.2224,
      "step": 1897
    },
    {
      "epoch": 1.972972972972973,
      "grad_norm": 3.9583427906036377,
      "learning_rate": 6.72903672903673e-06,
      "loss": 0.1476,
      "step": 1898
    },
    {
      "epoch": 1.974012474012474,
      "grad_norm": 2.645249605178833,
      "learning_rate": 6.726726726726728e-06,
      "loss": 0.1043,
      "step": 1899
    },
    {
      "epoch": 1.975051975051975,
      "grad_norm": 0.6136985421180725,
      "learning_rate": 6.724416724416725e-06,
      "loss": 0.0089,
      "step": 1900
    },
    {
      "epoch": 1.976091476091476,
      "grad_norm": 10.764155387878418,
      "learning_rate": 6.722106722106723e-06,
      "loss": 0.6656,
      "step": 1901
    },
    {
      "epoch": 1.9771309771309773,
      "grad_norm": 1.2077161073684692,
      "learning_rate": 6.719796719796721e-06,
      "loss": 0.0225,
      "step": 1902
    },
    {
      "epoch": 1.9781704781704783,
      "grad_norm": 7.769934177398682,
      "learning_rate": 6.7174867174867175e-06,
      "loss": 0.2133,
      "step": 1903
    },
    {
      "epoch": 1.9792099792099793,
      "grad_norm": 10.644739151000977,
      "learning_rate": 6.715176715176716e-06,
      "loss": 1.0806,
      "step": 1904
    },
    {
      "epoch": 1.9802494802494803,
      "grad_norm": 4.899578094482422,
      "learning_rate": 6.712866712866714e-06,
      "loss": 0.3573,
      "step": 1905
    },
    {
      "epoch": 1.9812889812889813,
      "grad_norm": 8.201702117919922,
      "learning_rate": 6.710556710556711e-06,
      "loss": 0.2393,
      "step": 1906
    },
    {
      "epoch": 1.9823284823284824,
      "grad_norm": 2.221436023712158,
      "learning_rate": 6.7082467082467085e-06,
      "loss": 0.0599,
      "step": 1907
    },
    {
      "epoch": 1.9833679833679834,
      "grad_norm": 1.037449836730957,
      "learning_rate": 6.705936705936707e-06,
      "loss": 0.0224,
      "step": 1908
    },
    {
      "epoch": 1.9844074844074844,
      "grad_norm": 0.08712287992238998,
      "learning_rate": 6.703626703626704e-06,
      "loss": 0.0016,
      "step": 1909
    },
    {
      "epoch": 1.9854469854469854,
      "grad_norm": 3.80332350730896,
      "learning_rate": 6.701316701316701e-06,
      "loss": 0.1323,
      "step": 1910
    },
    {
      "epoch": 1.9864864864864864,
      "grad_norm": 0.1044154092669487,
      "learning_rate": 6.6990066990067e-06,
      "loss": 0.0022,
      "step": 1911
    },
    {
      "epoch": 1.9875259875259874,
      "grad_norm": 0.1784604787826538,
      "learning_rate": 6.696696696696697e-06,
      "loss": 0.0067,
      "step": 1912
    },
    {
      "epoch": 1.9885654885654884,
      "grad_norm": 0.02075844444334507,
      "learning_rate": 6.694386694386695e-06,
      "loss": 0.0004,
      "step": 1913
    },
    {
      "epoch": 1.9896049896049897,
      "grad_norm": 10.95056438446045,
      "learning_rate": 6.6920766920766925e-06,
      "loss": 1.0339,
      "step": 1914
    },
    {
      "epoch": 1.9906444906444907,
      "grad_norm": 12.065203666687012,
      "learning_rate": 6.68976668976669e-06,
      "loss": 1.3071,
      "step": 1915
    },
    {
      "epoch": 1.9916839916839917,
      "grad_norm": 7.796991348266602,
      "learning_rate": 6.687456687456688e-06,
      "loss": 0.4829,
      "step": 1916
    },
    {
      "epoch": 1.9927234927234927,
      "grad_norm": 6.960740566253662,
      "learning_rate": 6.685146685146686e-06,
      "loss": 0.1915,
      "step": 1917
    },
    {
      "epoch": 1.993762993762994,
      "grad_norm": 11.084587097167969,
      "learning_rate": 6.682836682836683e-06,
      "loss": 1.2358,
      "step": 1918
    },
    {
      "epoch": 1.994802494802495,
      "grad_norm": 4.746704578399658,
      "learning_rate": 6.680526680526681e-06,
      "loss": 0.3151,
      "step": 1919
    },
    {
      "epoch": 1.995841995841996,
      "grad_norm": 8.840492248535156,
      "learning_rate": 6.678216678216679e-06,
      "loss": 0.8359,
      "step": 1920
    },
    {
      "epoch": 1.996881496881497,
      "grad_norm": 2.5447044372558594,
      "learning_rate": 6.675906675906676e-06,
      "loss": 0.0656,
      "step": 1921
    },
    {
      "epoch": 1.997920997920998,
      "grad_norm": 0.010701720602810383,
      "learning_rate": 6.673596673596674e-06,
      "loss": 0.0002,
      "step": 1922
    },
    {
      "epoch": 1.998960498960499,
      "grad_norm": 9.35302448272705,
      "learning_rate": 6.671286671286672e-06,
      "loss": 0.5405,
      "step": 1923
    },
    {
      "epoch": 2.0,
      "grad_norm": 31.449922561645508,
      "learning_rate": 6.66897666897667e-06,
      "loss": 1.1776,
      "step": 1924
    },
    {
      "epoch": 2.001039501039501,
      "grad_norm": 14.181282043457031,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.1872,
      "step": 1925
    },
    {
      "epoch": 2.002079002079002,
      "grad_norm": 4.408305644989014,
      "learning_rate": 6.664356664356665e-06,
      "loss": 0.2631,
      "step": 1926
    },
    {
      "epoch": 2.003118503118503,
      "grad_norm": 1.584025502204895,
      "learning_rate": 6.662046662046663e-06,
      "loss": 0.0318,
      "step": 1927
    },
    {
      "epoch": 2.004158004158004,
      "grad_norm": 0.3120782673358917,
      "learning_rate": 6.659736659736661e-06,
      "loss": 0.0054,
      "step": 1928
    },
    {
      "epoch": 2.005197505197505,
      "grad_norm": 0.15407142043113708,
      "learning_rate": 6.657426657426658e-06,
      "loss": 0.0044,
      "step": 1929
    },
    {
      "epoch": 2.006237006237006,
      "grad_norm": 6.331607341766357,
      "learning_rate": 6.655116655116656e-06,
      "loss": 0.1093,
      "step": 1930
    },
    {
      "epoch": 2.007276507276507,
      "grad_norm": 6.542793273925781,
      "learning_rate": 6.652806652806654e-06,
      "loss": 0.2508,
      "step": 1931
    },
    {
      "epoch": 2.008316008316008,
      "grad_norm": 1.4102624654769897,
      "learning_rate": 6.650496650496651e-06,
      "loss": 0.0119,
      "step": 1932
    },
    {
      "epoch": 2.0093555093555096,
      "grad_norm": 0.21904535591602325,
      "learning_rate": 6.648186648186649e-06,
      "loss": 0.0042,
      "step": 1933
    },
    {
      "epoch": 2.0103950103950106,
      "grad_norm": 0.6024210453033447,
      "learning_rate": 6.645876645876647e-06,
      "loss": 0.0171,
      "step": 1934
    },
    {
      "epoch": 2.0114345114345116,
      "grad_norm": 3.4335834980010986,
      "learning_rate": 6.643566643566644e-06,
      "loss": 0.0419,
      "step": 1935
    },
    {
      "epoch": 2.0124740124740126,
      "grad_norm": 25.000988006591797,
      "learning_rate": 6.641256641256642e-06,
      "loss": 0.9182,
      "step": 1936
    },
    {
      "epoch": 2.0135135135135136,
      "grad_norm": 3.822493314743042,
      "learning_rate": 6.63894663894664e-06,
      "loss": 0.2063,
      "step": 1937
    },
    {
      "epoch": 2.0145530145530146,
      "grad_norm": 0.0064015756361186504,
      "learning_rate": 6.636636636636637e-06,
      "loss": 0.0001,
      "step": 1938
    },
    {
      "epoch": 2.0155925155925156,
      "grad_norm": 2.934967279434204,
      "learning_rate": 6.6343266343266345e-06,
      "loss": 0.0684,
      "step": 1939
    },
    {
      "epoch": 2.0166320166320166,
      "grad_norm": 3.974259614944458,
      "learning_rate": 6.632016632016633e-06,
      "loss": 0.0803,
      "step": 1940
    },
    {
      "epoch": 2.0176715176715176,
      "grad_norm": 8.846273422241211,
      "learning_rate": 6.62970662970663e-06,
      "loss": 0.2658,
      "step": 1941
    },
    {
      "epoch": 2.0187110187110187,
      "grad_norm": 0.822585940361023,
      "learning_rate": 6.627396627396628e-06,
      "loss": 0.0143,
      "step": 1942
    },
    {
      "epoch": 2.0197505197505197,
      "grad_norm": 1.864228367805481,
      "learning_rate": 6.625086625086626e-06,
      "loss": 0.0349,
      "step": 1943
    },
    {
      "epoch": 2.0207900207900207,
      "grad_norm": 0.940016508102417,
      "learning_rate": 6.622776622776623e-06,
      "loss": 0.0192,
      "step": 1944
    },
    {
      "epoch": 2.0218295218295217,
      "grad_norm": 5.850529193878174,
      "learning_rate": 6.620466620466621e-06,
      "loss": 0.3462,
      "step": 1945
    },
    {
      "epoch": 2.0228690228690227,
      "grad_norm": 3.6243691444396973,
      "learning_rate": 6.618156618156619e-06,
      "loss": 0.0886,
      "step": 1946
    },
    {
      "epoch": 2.0239085239085237,
      "grad_norm": 0.3001081645488739,
      "learning_rate": 6.615846615846616e-06,
      "loss": 0.0076,
      "step": 1947
    },
    {
      "epoch": 2.024948024948025,
      "grad_norm": 4.391566276550293,
      "learning_rate": 6.613536613536614e-06,
      "loss": 0.125,
      "step": 1948
    },
    {
      "epoch": 2.025987525987526,
      "grad_norm": 0.49894315004348755,
      "learning_rate": 6.611226611226612e-06,
      "loss": 0.0112,
      "step": 1949
    },
    {
      "epoch": 2.027027027027027,
      "grad_norm": 8.597855567932129,
      "learning_rate": 6.608916608916609e-06,
      "loss": 0.4718,
      "step": 1950
    },
    {
      "epoch": 2.028066528066528,
      "grad_norm": 7.421142101287842,
      "learning_rate": 6.606606606606607e-06,
      "loss": 0.1622,
      "step": 1951
    },
    {
      "epoch": 2.029106029106029,
      "grad_norm": 0.04691620543599129,
      "learning_rate": 6.604296604296605e-06,
      "loss": 0.0009,
      "step": 1952
    },
    {
      "epoch": 2.0301455301455302,
      "grad_norm": 0.14164651930332184,
      "learning_rate": 6.601986601986603e-06,
      "loss": 0.0031,
      "step": 1953
    },
    {
      "epoch": 2.0311850311850312,
      "grad_norm": 1.1138488054275513,
      "learning_rate": 6.5996765996766e-06,
      "loss": 0.0305,
      "step": 1954
    },
    {
      "epoch": 2.0322245322245323,
      "grad_norm": 0.37105488777160645,
      "learning_rate": 6.597366597366598e-06,
      "loss": 0.0055,
      "step": 1955
    },
    {
      "epoch": 2.0332640332640333,
      "grad_norm": 6.09819221496582,
      "learning_rate": 6.595056595056596e-06,
      "loss": 0.3586,
      "step": 1956
    },
    {
      "epoch": 2.0343035343035343,
      "grad_norm": 0.18604886531829834,
      "learning_rate": 6.592746592746593e-06,
      "loss": 0.0034,
      "step": 1957
    },
    {
      "epoch": 2.0353430353430353,
      "grad_norm": 5.702450752258301,
      "learning_rate": 6.590436590436591e-06,
      "loss": 0.1071,
      "step": 1958
    },
    {
      "epoch": 2.0363825363825363,
      "grad_norm": 0.5846487283706665,
      "learning_rate": 6.588126588126589e-06,
      "loss": 0.0127,
      "step": 1959
    },
    {
      "epoch": 2.0374220374220373,
      "grad_norm": 11.294425964355469,
      "learning_rate": 6.585816585816586e-06,
      "loss": 0.682,
      "step": 1960
    },
    {
      "epoch": 2.0384615384615383,
      "grad_norm": 0.3569445013999939,
      "learning_rate": 6.583506583506584e-06,
      "loss": 0.0099,
      "step": 1961
    },
    {
      "epoch": 2.0395010395010393,
      "grad_norm": 4.077264785766602,
      "learning_rate": 6.581196581196582e-06,
      "loss": 0.1448,
      "step": 1962
    },
    {
      "epoch": 2.0405405405405403,
      "grad_norm": 3.6174886226654053,
      "learning_rate": 6.578886578886579e-06,
      "loss": 0.1382,
      "step": 1963
    },
    {
      "epoch": 2.0415800415800414,
      "grad_norm": 0.5434964299201965,
      "learning_rate": 6.5765765765765775e-06,
      "loss": 0.0201,
      "step": 1964
    },
    {
      "epoch": 2.042619542619543,
      "grad_norm": 0.6819065809249878,
      "learning_rate": 6.574266574266575e-06,
      "loss": 0.0219,
      "step": 1965
    },
    {
      "epoch": 2.043659043659044,
      "grad_norm": 14.37651538848877,
      "learning_rate": 6.571956571956572e-06,
      "loss": 1.741,
      "step": 1966
    },
    {
      "epoch": 2.044698544698545,
      "grad_norm": 8.81903076171875,
      "learning_rate": 6.56964656964657e-06,
      "loss": 0.2543,
      "step": 1967
    },
    {
      "epoch": 2.045738045738046,
      "grad_norm": 6.487217903137207,
      "learning_rate": 6.567336567336568e-06,
      "loss": 0.2681,
      "step": 1968
    },
    {
      "epoch": 2.046777546777547,
      "grad_norm": 9.601208686828613,
      "learning_rate": 6.565026565026565e-06,
      "loss": 0.5295,
      "step": 1969
    },
    {
      "epoch": 2.047817047817048,
      "grad_norm": 14.670531272888184,
      "learning_rate": 6.562716562716563e-06,
      "loss": 2.0842,
      "step": 1970
    },
    {
      "epoch": 2.048856548856549,
      "grad_norm": 11.62336540222168,
      "learning_rate": 6.560406560406561e-06,
      "loss": 0.4462,
      "step": 1971
    },
    {
      "epoch": 2.04989604989605,
      "grad_norm": 0.23016346991062164,
      "learning_rate": 6.558096558096559e-06,
      "loss": 0.0071,
      "step": 1972
    },
    {
      "epoch": 2.050935550935551,
      "grad_norm": 5.836270332336426,
      "learning_rate": 6.555786555786556e-06,
      "loss": 0.3069,
      "step": 1973
    },
    {
      "epoch": 2.051975051975052,
      "grad_norm": 0.14788639545440674,
      "learning_rate": 6.553476553476554e-06,
      "loss": 0.003,
      "step": 1974
    },
    {
      "epoch": 2.053014553014553,
      "grad_norm": 5.114451885223389,
      "learning_rate": 6.551166551166552e-06,
      "loss": 0.3132,
      "step": 1975
    },
    {
      "epoch": 2.054054054054054,
      "grad_norm": 0.022919783368706703,
      "learning_rate": 6.548856548856549e-06,
      "loss": 0.0006,
      "step": 1976
    },
    {
      "epoch": 2.055093555093555,
      "grad_norm": 10.959732055664062,
      "learning_rate": 6.546546546546547e-06,
      "loss": 0.6432,
      "step": 1977
    },
    {
      "epoch": 2.056133056133056,
      "grad_norm": 0.09819675981998444,
      "learning_rate": 6.544236544236545e-06,
      "loss": 0.002,
      "step": 1978
    },
    {
      "epoch": 2.057172557172557,
      "grad_norm": 17.36860466003418,
      "learning_rate": 6.541926541926542e-06,
      "loss": 1.5427,
      "step": 1979
    },
    {
      "epoch": 2.0582120582120584,
      "grad_norm": 2.121630907058716,
      "learning_rate": 6.53961653961654e-06,
      "loss": 0.0472,
      "step": 1980
    },
    {
      "epoch": 2.0592515592515594,
      "grad_norm": 12.962278366088867,
      "learning_rate": 6.537306537306538e-06,
      "loss": 1.041,
      "step": 1981
    },
    {
      "epoch": 2.0602910602910605,
      "grad_norm": 8.771272659301758,
      "learning_rate": 6.534996534996536e-06,
      "loss": 0.389,
      "step": 1982
    },
    {
      "epoch": 2.0613305613305615,
      "grad_norm": 8.74985408782959,
      "learning_rate": 6.532686532686533e-06,
      "loss": 0.5721,
      "step": 1983
    },
    {
      "epoch": 2.0623700623700625,
      "grad_norm": 5.376676559448242,
      "learning_rate": 6.530376530376531e-06,
      "loss": 0.2728,
      "step": 1984
    },
    {
      "epoch": 2.0634095634095635,
      "grad_norm": 6.670430660247803,
      "learning_rate": 6.528066528066529e-06,
      "loss": 0.6577,
      "step": 1985
    },
    {
      "epoch": 2.0644490644490645,
      "grad_norm": 6.218632698059082,
      "learning_rate": 6.525756525756526e-06,
      "loss": 0.1806,
      "step": 1986
    },
    {
      "epoch": 2.0654885654885655,
      "grad_norm": 0.47197651863098145,
      "learning_rate": 6.523446523446524e-06,
      "loss": 0.0076,
      "step": 1987
    },
    {
      "epoch": 2.0665280665280665,
      "grad_norm": 3.8785648345947266,
      "learning_rate": 6.521136521136522e-06,
      "loss": 0.0395,
      "step": 1988
    },
    {
      "epoch": 2.0675675675675675,
      "grad_norm": 0.2819497883319855,
      "learning_rate": 6.5188265188265195e-06,
      "loss": 0.0078,
      "step": 1989
    },
    {
      "epoch": 2.0686070686070686,
      "grad_norm": 2.064453125,
      "learning_rate": 6.516516516516517e-06,
      "loss": 0.0416,
      "step": 1990
    },
    {
      "epoch": 2.0696465696465696,
      "grad_norm": 0.17872098088264465,
      "learning_rate": 6.514206514206515e-06,
      "loss": 0.007,
      "step": 1991
    },
    {
      "epoch": 2.0706860706860706,
      "grad_norm": 4.090879440307617,
      "learning_rate": 6.511896511896512e-06,
      "loss": 0.1246,
      "step": 1992
    },
    {
      "epoch": 2.0717255717255716,
      "grad_norm": 9.653943061828613,
      "learning_rate": 6.50958650958651e-06,
      "loss": 1.0796,
      "step": 1993
    },
    {
      "epoch": 2.0727650727650726,
      "grad_norm": 5.251039981842041,
      "learning_rate": 6.507276507276508e-06,
      "loss": 0.2909,
      "step": 1994
    },
    {
      "epoch": 2.0738045738045736,
      "grad_norm": 1.227087378501892,
      "learning_rate": 6.504966504966505e-06,
      "loss": 0.049,
      "step": 1995
    },
    {
      "epoch": 2.0748440748440746,
      "grad_norm": 8.762991905212402,
      "learning_rate": 6.5026565026565035e-06,
      "loss": 0.8551,
      "step": 1996
    },
    {
      "epoch": 2.075883575883576,
      "grad_norm": 2.4482312202453613,
      "learning_rate": 6.500346500346501e-06,
      "loss": 0.0475,
      "step": 1997
    },
    {
      "epoch": 2.076923076923077,
      "grad_norm": 9.400773048400879,
      "learning_rate": 6.498036498036498e-06,
      "loss": 0.9039,
      "step": 1998
    },
    {
      "epoch": 2.077962577962578,
      "grad_norm": 9.062102317810059,
      "learning_rate": 6.495726495726496e-06,
      "loss": 0.4474,
      "step": 1999
    },
    {
      "epoch": 2.079002079002079,
      "grad_norm": 6.728084564208984,
      "learning_rate": 6.4934164934164945e-06,
      "loss": 0.417,
      "step": 2000
    },
    {
      "epoch": 2.08004158004158,
      "grad_norm": 5.888612270355225,
      "learning_rate": 6.491106491106491e-06,
      "loss": 0.1305,
      "step": 2001
    },
    {
      "epoch": 2.081081081081081,
      "grad_norm": 1.8210113048553467,
      "learning_rate": 6.488796488796489e-06,
      "loss": 0.0545,
      "step": 2002
    },
    {
      "epoch": 2.082120582120582,
      "grad_norm": 3.953402280807495,
      "learning_rate": 6.486486486486487e-06,
      "loss": 0.156,
      "step": 2003
    },
    {
      "epoch": 2.083160083160083,
      "grad_norm": 1.5353587865829468,
      "learning_rate": 6.484176484176484e-06,
      "loss": 0.037,
      "step": 2004
    },
    {
      "epoch": 2.084199584199584,
      "grad_norm": 6.91809606552124,
      "learning_rate": 6.481866481866482e-06,
      "loss": 0.8773,
      "step": 2005
    },
    {
      "epoch": 2.085239085239085,
      "grad_norm": 4.051370620727539,
      "learning_rate": 6.47955647955648e-06,
      "loss": 0.2595,
      "step": 2006
    },
    {
      "epoch": 2.086278586278586,
      "grad_norm": 10.553234100341797,
      "learning_rate": 6.4772464772464785e-06,
      "loss": 0.7355,
      "step": 2007
    },
    {
      "epoch": 2.087318087318087,
      "grad_norm": 0.09537404030561447,
      "learning_rate": 6.474936474936475e-06,
      "loss": 0.0029,
      "step": 2008
    },
    {
      "epoch": 2.0883575883575882,
      "grad_norm": 11.784072875976562,
      "learning_rate": 6.472626472626473e-06,
      "loss": 0.1916,
      "step": 2009
    },
    {
      "epoch": 2.0893970893970892,
      "grad_norm": 1.224432349205017,
      "learning_rate": 6.470316470316471e-06,
      "loss": 0.0435,
      "step": 2010
    },
    {
      "epoch": 2.0904365904365902,
      "grad_norm": 3.884338617324829,
      "learning_rate": 6.468006468006468e-06,
      "loss": 0.0429,
      "step": 2011
    },
    {
      "epoch": 2.0914760914760917,
      "grad_norm": 2.474146842956543,
      "learning_rate": 6.465696465696466e-06,
      "loss": 0.062,
      "step": 2012
    },
    {
      "epoch": 2.0925155925155927,
      "grad_norm": 1.0791113376617432,
      "learning_rate": 6.463386463386464e-06,
      "loss": 0.0222,
      "step": 2013
    },
    {
      "epoch": 2.0935550935550937,
      "grad_norm": 9.401138305664062,
      "learning_rate": 6.461076461076462e-06,
      "loss": 0.6004,
      "step": 2014
    },
    {
      "epoch": 2.0945945945945947,
      "grad_norm": 1.2327183485031128,
      "learning_rate": 6.458766458766459e-06,
      "loss": 0.0233,
      "step": 2015
    },
    {
      "epoch": 2.0956340956340958,
      "grad_norm": 0.08808104693889618,
      "learning_rate": 6.456456456456457e-06,
      "loss": 0.0018,
      "step": 2016
    },
    {
      "epoch": 2.0966735966735968,
      "grad_norm": 4.59703254699707,
      "learning_rate": 6.454146454146455e-06,
      "loss": 0.0998,
      "step": 2017
    },
    {
      "epoch": 2.0977130977130978,
      "grad_norm": 8.11798095703125,
      "learning_rate": 6.451836451836453e-06,
      "loss": 0.5019,
      "step": 2018
    },
    {
      "epoch": 2.098752598752599,
      "grad_norm": 2.0150322914123535,
      "learning_rate": 6.44952644952645e-06,
      "loss": 0.0466,
      "step": 2019
    },
    {
      "epoch": 2.0997920997921,
      "grad_norm": 4.207459926605225,
      "learning_rate": 6.447216447216448e-06,
      "loss": 0.1545,
      "step": 2020
    },
    {
      "epoch": 2.100831600831601,
      "grad_norm": 2.6554224491119385,
      "learning_rate": 6.4449064449064455e-06,
      "loss": 0.0605,
      "step": 2021
    },
    {
      "epoch": 2.101871101871102,
      "grad_norm": 1.8698917627334595,
      "learning_rate": 6.442596442596443e-06,
      "loss": 0.0471,
      "step": 2022
    },
    {
      "epoch": 2.102910602910603,
      "grad_norm": 6.580633640289307,
      "learning_rate": 6.440286440286441e-06,
      "loss": 0.1459,
      "step": 2023
    },
    {
      "epoch": 2.103950103950104,
      "grad_norm": 8.103753089904785,
      "learning_rate": 6.437976437976438e-06,
      "loss": 0.2792,
      "step": 2024
    },
    {
      "epoch": 2.104989604989605,
      "grad_norm": 0.27463245391845703,
      "learning_rate": 6.435666435666437e-06,
      "loss": 0.0083,
      "step": 2025
    },
    {
      "epoch": 2.106029106029106,
      "grad_norm": 1.2727649211883545,
      "learning_rate": 6.433356433356434e-06,
      "loss": 0.0291,
      "step": 2026
    },
    {
      "epoch": 2.107068607068607,
      "grad_norm": 8.371627807617188,
      "learning_rate": 6.431046431046431e-06,
      "loss": 0.2413,
      "step": 2027
    },
    {
      "epoch": 2.108108108108108,
      "grad_norm": 7.761322021484375,
      "learning_rate": 6.4287364287364295e-06,
      "loss": 0.474,
      "step": 2028
    },
    {
      "epoch": 2.1091476091476093,
      "grad_norm": 3.2681360244750977,
      "learning_rate": 6.426426426426427e-06,
      "loss": 0.1832,
      "step": 2029
    },
    {
      "epoch": 2.1101871101871104,
      "grad_norm": 4.478689193725586,
      "learning_rate": 6.424116424116424e-06,
      "loss": 0.3446,
      "step": 2030
    },
    {
      "epoch": 2.1112266112266114,
      "grad_norm": 7.7292985916137695,
      "learning_rate": 6.421806421806422e-06,
      "loss": 0.2425,
      "step": 2031
    },
    {
      "epoch": 2.1122661122661124,
      "grad_norm": 3.809605598449707,
      "learning_rate": 6.4194964194964205e-06,
      "loss": 0.0945,
      "step": 2032
    },
    {
      "epoch": 2.1133056133056134,
      "grad_norm": 0.0223262719810009,
      "learning_rate": 6.417186417186417e-06,
      "loss": 0.0005,
      "step": 2033
    },
    {
      "epoch": 2.1143451143451144,
      "grad_norm": 3.581709861755371,
      "learning_rate": 6.414876414876415e-06,
      "loss": 0.0982,
      "step": 2034
    },
    {
      "epoch": 2.1153846153846154,
      "grad_norm": 1.2276995182037354,
      "learning_rate": 6.412566412566413e-06,
      "loss": 0.02,
      "step": 2035
    },
    {
      "epoch": 2.1164241164241164,
      "grad_norm": 11.483390808105469,
      "learning_rate": 6.410256410256412e-06,
      "loss": 1.2713,
      "step": 2036
    },
    {
      "epoch": 2.1174636174636174,
      "grad_norm": 0.03083518147468567,
      "learning_rate": 6.407946407946408e-06,
      "loss": 0.0009,
      "step": 2037
    },
    {
      "epoch": 2.1185031185031185,
      "grad_norm": 6.554765701293945,
      "learning_rate": 6.405636405636406e-06,
      "loss": 0.3977,
      "step": 2038
    },
    {
      "epoch": 2.1195426195426195,
      "grad_norm": 1.5237971544265747,
      "learning_rate": 6.4033264033264045e-06,
      "loss": 0.0328,
      "step": 2039
    },
    {
      "epoch": 2.1205821205821205,
      "grad_norm": 9.13638687133789,
      "learning_rate": 6.401016401016401e-06,
      "loss": 0.679,
      "step": 2040
    },
    {
      "epoch": 2.1216216216216215,
      "grad_norm": 6.517833709716797,
      "learning_rate": 6.398706398706399e-06,
      "loss": 0.2226,
      "step": 2041
    },
    {
      "epoch": 2.1226611226611225,
      "grad_norm": 0.15088814496994019,
      "learning_rate": 6.396396396396397e-06,
      "loss": 0.0047,
      "step": 2042
    },
    {
      "epoch": 2.1237006237006235,
      "grad_norm": 7.6674723625183105,
      "learning_rate": 6.394086394086395e-06,
      "loss": 0.4219,
      "step": 2043
    },
    {
      "epoch": 2.124740124740125,
      "grad_norm": 11.87431526184082,
      "learning_rate": 6.391776391776392e-06,
      "loss": 0.0741,
      "step": 2044
    },
    {
      "epoch": 2.125779625779626,
      "grad_norm": 9.523569107055664,
      "learning_rate": 6.38946638946639e-06,
      "loss": 0.6066,
      "step": 2045
    },
    {
      "epoch": 2.126819126819127,
      "grad_norm": 7.635879039764404,
      "learning_rate": 6.387156387156388e-06,
      "loss": 0.6455,
      "step": 2046
    },
    {
      "epoch": 2.127858627858628,
      "grad_norm": 1.5417159795761108,
      "learning_rate": 6.384846384846385e-06,
      "loss": 0.0354,
      "step": 2047
    },
    {
      "epoch": 2.128898128898129,
      "grad_norm": 5.412545204162598,
      "learning_rate": 6.382536382536383e-06,
      "loss": 0.1233,
      "step": 2048
    },
    {
      "epoch": 2.12993762993763,
      "grad_norm": 3.4599087238311768,
      "learning_rate": 6.3802263802263805e-06,
      "loss": 0.074,
      "step": 2049
    },
    {
      "epoch": 2.130977130977131,
      "grad_norm": 0.6040581464767456,
      "learning_rate": 6.377916377916379e-06,
      "loss": 0.0182,
      "step": 2050
    },
    {
      "epoch": 2.132016632016632,
      "grad_norm": 7.773816108703613,
      "learning_rate": 6.375606375606376e-06,
      "loss": 0.5669,
      "step": 2051
    },
    {
      "epoch": 2.133056133056133,
      "grad_norm": 0.4341773986816406,
      "learning_rate": 6.373296373296373e-06,
      "loss": 0.0112,
      "step": 2052
    },
    {
      "epoch": 2.134095634095634,
      "grad_norm": NaN,
      "learning_rate": 6.3709863709863715e-06,
      "loss": 0.2966,
      "step": 2053
    },
    {
      "epoch": 2.135135135135135,
      "grad_norm": 2.1740145683288574,
      "learning_rate": 6.36867636867637e-06,
      "loss": 0.0534,
      "step": 2054
    },
    {
      "epoch": 2.136174636174636,
      "grad_norm": 0.956126868724823,
      "learning_rate": 6.366366366366366e-06,
      "loss": 0.0263,
      "step": 2055
    },
    {
      "epoch": 2.137214137214137,
      "grad_norm": 2.815540075302124,
      "learning_rate": 6.364056364056364e-06,
      "loss": 0.0602,
      "step": 2056
    },
    {
      "epoch": 2.138253638253638,
      "grad_norm": 5.632743835449219,
      "learning_rate": 6.361746361746363e-06,
      "loss": 0.1353,
      "step": 2057
    },
    {
      "epoch": 2.139293139293139,
      "grad_norm": 7.705263614654541,
      "learning_rate": 6.359436359436359e-06,
      "loss": 1.0678,
      "step": 2058
    },
    {
      "epoch": 2.14033264033264,
      "grad_norm": 10.012617111206055,
      "learning_rate": 6.357126357126357e-06,
      "loss": 0.8383,
      "step": 2059
    },
    {
      "epoch": 2.141372141372141,
      "grad_norm": 11.808959007263184,
      "learning_rate": 6.3548163548163555e-06,
      "loss": 0.9213,
      "step": 2060
    },
    {
      "epoch": 2.1424116424116426,
      "grad_norm": 6.768054008483887,
      "learning_rate": 6.352506352506354e-06,
      "loss": 0.7591,
      "step": 2061
    },
    {
      "epoch": 2.1434511434511436,
      "grad_norm": 8.046747207641602,
      "learning_rate": 6.35019635019635e-06,
      "loss": 0.8153,
      "step": 2062
    },
    {
      "epoch": 2.1444906444906446,
      "grad_norm": 4.566589832305908,
      "learning_rate": 6.347886347886348e-06,
      "loss": 0.3171,
      "step": 2063
    },
    {
      "epoch": 2.1455301455301456,
      "grad_norm": 7.478272914886475,
      "learning_rate": 6.3455763455763465e-06,
      "loss": 0.5232,
      "step": 2064
    },
    {
      "epoch": 2.1465696465696467,
      "grad_norm": 9.267080307006836,
      "learning_rate": 6.343266343266343e-06,
      "loss": 0.2926,
      "step": 2065
    },
    {
      "epoch": 2.1476091476091477,
      "grad_norm": 12.988465309143066,
      "learning_rate": 6.340956340956341e-06,
      "loss": 0.3343,
      "step": 2066
    },
    {
      "epoch": 2.1486486486486487,
      "grad_norm": 7.443843841552734,
      "learning_rate": 6.3386463386463394e-06,
      "loss": 0.4648,
      "step": 2067
    },
    {
      "epoch": 2.1496881496881497,
      "grad_norm": 1.7085425853729248,
      "learning_rate": 6.336336336336338e-06,
      "loss": 0.0593,
      "step": 2068
    },
    {
      "epoch": 2.1507276507276507,
      "grad_norm": 1.6968820095062256,
      "learning_rate": 6.334026334026334e-06,
      "loss": 0.0684,
      "step": 2069
    },
    {
      "epoch": 2.1517671517671517,
      "grad_norm": 2.399233102798462,
      "learning_rate": 6.331716331716332e-06,
      "loss": 0.0861,
      "step": 2070
    },
    {
      "epoch": 2.1528066528066527,
      "grad_norm": 4.714275360107422,
      "learning_rate": 6.3294063294063305e-06,
      "loss": 0.2724,
      "step": 2071
    },
    {
      "epoch": 2.1538461538461537,
      "grad_norm": 8.669092178344727,
      "learning_rate": 6.327096327096328e-06,
      "loss": 0.5226,
      "step": 2072
    },
    {
      "epoch": 2.1548856548856548,
      "grad_norm": 0.6544547080993652,
      "learning_rate": 6.324786324786325e-06,
      "loss": 0.0181,
      "step": 2073
    },
    {
      "epoch": 2.1559251559251558,
      "grad_norm": 5.61388635635376,
      "learning_rate": 6.322476322476323e-06,
      "loss": 0.3196,
      "step": 2074
    },
    {
      "epoch": 2.156964656964657,
      "grad_norm": 11.46261215209961,
      "learning_rate": 6.320166320166321e-06,
      "loss": 0.6334,
      "step": 2075
    },
    {
      "epoch": 2.1580041580041582,
      "grad_norm": 1.828363060951233,
      "learning_rate": 6.317856317856318e-06,
      "loss": 0.062,
      "step": 2076
    },
    {
      "epoch": 2.1590436590436592,
      "grad_norm": 4.612522125244141,
      "learning_rate": 6.315546315546316e-06,
      "loss": 0.2484,
      "step": 2077
    },
    {
      "epoch": 2.1600831600831603,
      "grad_norm": 3.598356246948242,
      "learning_rate": 6.313236313236314e-06,
      "loss": 0.0767,
      "step": 2078
    },
    {
      "epoch": 2.1611226611226613,
      "grad_norm": 7.803494453430176,
      "learning_rate": 6.310926310926312e-06,
      "loss": 0.8641,
      "step": 2079
    },
    {
      "epoch": 2.1621621621621623,
      "grad_norm": 4.039019584655762,
      "learning_rate": 6.308616308616309e-06,
      "loss": 0.3692,
      "step": 2080
    },
    {
      "epoch": 2.1632016632016633,
      "grad_norm": 0.3767847716808319,
      "learning_rate": 6.3063063063063065e-06,
      "loss": 0.0117,
      "step": 2081
    },
    {
      "epoch": 2.1642411642411643,
      "grad_norm": 1.3667638301849365,
      "learning_rate": 6.303996303996305e-06,
      "loss": 0.0471,
      "step": 2082
    },
    {
      "epoch": 2.1652806652806653,
      "grad_norm": 0.47995445132255554,
      "learning_rate": 6.301686301686302e-06,
      "loss": 0.0116,
      "step": 2083
    },
    {
      "epoch": 2.1663201663201663,
      "grad_norm": 1.2004308700561523,
      "learning_rate": 6.299376299376299e-06,
      "loss": 0.0399,
      "step": 2084
    },
    {
      "epoch": 2.1673596673596673,
      "grad_norm": 5.898972988128662,
      "learning_rate": 6.2970662970662975e-06,
      "loss": 0.1451,
      "step": 2085
    },
    {
      "epoch": 2.1683991683991684,
      "grad_norm": 7.3125386238098145,
      "learning_rate": 6.294756294756296e-06,
      "loss": 0.4661,
      "step": 2086
    },
    {
      "epoch": 2.1694386694386694,
      "grad_norm": 6.4429521560668945,
      "learning_rate": 6.292446292446292e-06,
      "loss": 0.2921,
      "step": 2087
    },
    {
      "epoch": 2.1704781704781704,
      "grad_norm": 1.5604861974716187,
      "learning_rate": 6.29013629013629e-06,
      "loss": 0.0699,
      "step": 2088
    },
    {
      "epoch": 2.1715176715176714,
      "grad_norm": 5.064148902893066,
      "learning_rate": 6.287826287826289e-06,
      "loss": 0.2257,
      "step": 2089
    },
    {
      "epoch": 2.1725571725571724,
      "grad_norm": 0.1482357233762741,
      "learning_rate": 6.285516285516287e-06,
      "loss": 0.0038,
      "step": 2090
    },
    {
      "epoch": 2.1735966735966734,
      "grad_norm": 0.17243508994579315,
      "learning_rate": 6.283206283206283e-06,
      "loss": 0.0035,
      "step": 2091
    },
    {
      "epoch": 2.1746361746361744,
      "grad_norm": 5.148295879364014,
      "learning_rate": 6.2808962808962815e-06,
      "loss": 0.1856,
      "step": 2092
    },
    {
      "epoch": 2.175675675675676,
      "grad_norm": 1.1720030307769775,
      "learning_rate": 6.27858627858628e-06,
      "loss": 0.0359,
      "step": 2093
    },
    {
      "epoch": 2.176715176715177,
      "grad_norm": 0.01272608246654272,
      "learning_rate": 6.276276276276276e-06,
      "loss": 0.0003,
      "step": 2094
    },
    {
      "epoch": 2.177754677754678,
      "grad_norm": 0.7603575587272644,
      "learning_rate": 6.273966273966274e-06,
      "loss": 0.0186,
      "step": 2095
    },
    {
      "epoch": 2.178794178794179,
      "grad_norm": 6.515388011932373,
      "learning_rate": 6.2716562716562726e-06,
      "loss": 0.331,
      "step": 2096
    },
    {
      "epoch": 2.17983367983368,
      "grad_norm": 8.437469482421875,
      "learning_rate": 6.269346269346271e-06,
      "loss": 0.7164,
      "step": 2097
    },
    {
      "epoch": 2.180873180873181,
      "grad_norm": 5.623338222503662,
      "learning_rate": 6.267036267036267e-06,
      "loss": 0.5556,
      "step": 2098
    },
    {
      "epoch": 2.181912681912682,
      "grad_norm": 8.537968635559082,
      "learning_rate": 6.2647262647262654e-06,
      "loss": 0.2938,
      "step": 2099
    },
    {
      "epoch": 2.182952182952183,
      "grad_norm": 0.11690367013216019,
      "learning_rate": 6.262416262416264e-06,
      "loss": 0.0028,
      "step": 2100
    },
    {
      "epoch": 2.183991683991684,
      "grad_norm": 7.875816345214844,
      "learning_rate": 6.26010626010626e-06,
      "loss": 0.7371,
      "step": 2101
    },
    {
      "epoch": 2.185031185031185,
      "grad_norm": 10.33857250213623,
      "learning_rate": 6.257796257796258e-06,
      "loss": 0.4181,
      "step": 2102
    },
    {
      "epoch": 2.186070686070686,
      "grad_norm": 7.2350006103515625,
      "learning_rate": 6.2554862554862565e-06,
      "loss": 0.2492,
      "step": 2103
    },
    {
      "epoch": 2.187110187110187,
      "grad_norm": 0.10823694616556168,
      "learning_rate": 6.253176253176254e-06,
      "loss": 0.0022,
      "step": 2104
    },
    {
      "epoch": 2.188149688149688,
      "grad_norm": 9.957201957702637,
      "learning_rate": 6.250866250866251e-06,
      "loss": 0.8395,
      "step": 2105
    },
    {
      "epoch": 2.189189189189189,
      "grad_norm": 5.966064453125,
      "learning_rate": 6.248556248556249e-06,
      "loss": 0.1804,
      "step": 2106
    },
    {
      "epoch": 2.19022869022869,
      "grad_norm": 10.796107292175293,
      "learning_rate": 6.246246246246247e-06,
      "loss": 0.8583,
      "step": 2107
    },
    {
      "epoch": 2.1912681912681915,
      "grad_norm": 2.4162845611572266,
      "learning_rate": 6.243936243936245e-06,
      "loss": 0.0993,
      "step": 2108
    },
    {
      "epoch": 2.1923076923076925,
      "grad_norm": 2.9660913944244385,
      "learning_rate": 6.241626241626242e-06,
      "loss": 0.1438,
      "step": 2109
    },
    {
      "epoch": 2.1933471933471935,
      "grad_norm": 0.12198910117149353,
      "learning_rate": 6.23931623931624e-06,
      "loss": 0.0041,
      "step": 2110
    },
    {
      "epoch": 2.1943866943866945,
      "grad_norm": 7.1431193351745605,
      "learning_rate": 6.237006237006238e-06,
      "loss": 0.6707,
      "step": 2111
    },
    {
      "epoch": 2.1954261954261955,
      "grad_norm": 8.095113754272461,
      "learning_rate": 6.234696234696235e-06,
      "loss": 0.8671,
      "step": 2112
    },
    {
      "epoch": 2.1964656964656966,
      "grad_norm": 0.2271644026041031,
      "learning_rate": 6.2323862323862325e-06,
      "loss": 0.0055,
      "step": 2113
    },
    {
      "epoch": 2.1975051975051976,
      "grad_norm": 2.605755567550659,
      "learning_rate": 6.230076230076231e-06,
      "loss": 0.0869,
      "step": 2114
    },
    {
      "epoch": 2.1985446985446986,
      "grad_norm": 0.5366362929344177,
      "learning_rate": 6.227766227766229e-06,
      "loss": 0.017,
      "step": 2115
    },
    {
      "epoch": 2.1995841995841996,
      "grad_norm": 3.6487009525299072,
      "learning_rate": 6.225456225456225e-06,
      "loss": 0.3753,
      "step": 2116
    },
    {
      "epoch": 2.2006237006237006,
      "grad_norm": 1.8559633493423462,
      "learning_rate": 6.2231462231462235e-06,
      "loss": 0.0677,
      "step": 2117
    },
    {
      "epoch": 2.2016632016632016,
      "grad_norm": 3.4091577529907227,
      "learning_rate": 6.220836220836222e-06,
      "loss": 0.1174,
      "step": 2118
    },
    {
      "epoch": 2.2027027027027026,
      "grad_norm": 4.887331962585449,
      "learning_rate": 6.218526218526218e-06,
      "loss": 0.1881,
      "step": 2119
    },
    {
      "epoch": 2.2037422037422036,
      "grad_norm": 6.882686614990234,
      "learning_rate": 6.2162162162162164e-06,
      "loss": 0.1934,
      "step": 2120
    },
    {
      "epoch": 2.2047817047817047,
      "grad_norm": 3.879831075668335,
      "learning_rate": 6.213906213906215e-06,
      "loss": 0.14,
      "step": 2121
    },
    {
      "epoch": 2.2058212058212057,
      "grad_norm": 0.22170592844486237,
      "learning_rate": 6.211596211596213e-06,
      "loss": 0.0072,
      "step": 2122
    },
    {
      "epoch": 2.2068607068607067,
      "grad_norm": 3.5209622383117676,
      "learning_rate": 6.209286209286209e-06,
      "loss": 0.2002,
      "step": 2123
    },
    {
      "epoch": 2.2079002079002077,
      "grad_norm": 0.3155515491962433,
      "learning_rate": 6.2069762069762075e-06,
      "loss": 0.0072,
      "step": 2124
    },
    {
      "epoch": 2.208939708939709,
      "grad_norm": 10.6986722946167,
      "learning_rate": 6.204666204666206e-06,
      "loss": 1.0673,
      "step": 2125
    },
    {
      "epoch": 2.20997920997921,
      "grad_norm": 1.0658669471740723,
      "learning_rate": 6.202356202356203e-06,
      "loss": 0.033,
      "step": 2126
    },
    {
      "epoch": 2.211018711018711,
      "grad_norm": 2.3950462341308594,
      "learning_rate": 6.2000462000462e-06,
      "loss": 0.039,
      "step": 2127
    },
    {
      "epoch": 2.212058212058212,
      "grad_norm": 0.15888486802577972,
      "learning_rate": 6.1977361977361986e-06,
      "loss": 0.0045,
      "step": 2128
    },
    {
      "epoch": 2.213097713097713,
      "grad_norm": 0.27664533257484436,
      "learning_rate": 6.195426195426196e-06,
      "loss": 0.0083,
      "step": 2129
    },
    {
      "epoch": 2.214137214137214,
      "grad_norm": 5.933618068695068,
      "learning_rate": 6.193116193116193e-06,
      "loss": 0.0743,
      "step": 2130
    },
    {
      "epoch": 2.215176715176715,
      "grad_norm": 9.662496566772461,
      "learning_rate": 6.1908061908061914e-06,
      "loss": 1.094,
      "step": 2131
    },
    {
      "epoch": 2.2162162162162162,
      "grad_norm": 3.5638606548309326,
      "learning_rate": 6.188496188496189e-06,
      "loss": 0.109,
      "step": 2132
    },
    {
      "epoch": 2.2172557172557172,
      "grad_norm": 6.295956134796143,
      "learning_rate": 6.186186186186187e-06,
      "loss": 0.1275,
      "step": 2133
    },
    {
      "epoch": 2.2182952182952183,
      "grad_norm": 3.0353546142578125,
      "learning_rate": 6.183876183876184e-06,
      "loss": 0.0632,
      "step": 2134
    },
    {
      "epoch": 2.2193347193347193,
      "grad_norm": 0.2624787986278534,
      "learning_rate": 6.181566181566182e-06,
      "loss": 0.0086,
      "step": 2135
    },
    {
      "epoch": 2.2203742203742203,
      "grad_norm": 9.579380989074707,
      "learning_rate": 6.17925617925618e-06,
      "loss": 0.3809,
      "step": 2136
    },
    {
      "epoch": 2.2214137214137213,
      "grad_norm": 0.9549159407615662,
      "learning_rate": 6.176946176946177e-06,
      "loss": 0.0227,
      "step": 2137
    },
    {
      "epoch": 2.2224532224532223,
      "grad_norm": 10.02344036102295,
      "learning_rate": 6.1746361746361745e-06,
      "loss": 0.8545,
      "step": 2138
    },
    {
      "epoch": 2.2234927234927233,
      "grad_norm": 7.736490726470947,
      "learning_rate": 6.172326172326173e-06,
      "loss": 0.5988,
      "step": 2139
    },
    {
      "epoch": 2.2245322245322248,
      "grad_norm": 0.9537568092346191,
      "learning_rate": 6.170016170016171e-06,
      "loss": 0.0187,
      "step": 2140
    },
    {
      "epoch": 2.225571725571726,
      "grad_norm": 10.918013572692871,
      "learning_rate": 6.167706167706167e-06,
      "loss": 1.3098,
      "step": 2141
    },
    {
      "epoch": 2.226611226611227,
      "grad_norm": 3.5731489658355713,
      "learning_rate": 6.165396165396166e-06,
      "loss": 0.0904,
      "step": 2142
    },
    {
      "epoch": 2.227650727650728,
      "grad_norm": 1.4764057397842407,
      "learning_rate": 6.163086163086164e-06,
      "loss": 0.0278,
      "step": 2143
    },
    {
      "epoch": 2.228690228690229,
      "grad_norm": 1.8778538703918457,
      "learning_rate": 6.160776160776162e-06,
      "loss": 0.0557,
      "step": 2144
    },
    {
      "epoch": 2.22972972972973,
      "grad_norm": 4.918324947357178,
      "learning_rate": 6.1584661584661585e-06,
      "loss": 0.1372,
      "step": 2145
    },
    {
      "epoch": 2.230769230769231,
      "grad_norm": 2.1157333850860596,
      "learning_rate": 6.156156156156157e-06,
      "loss": 0.0588,
      "step": 2146
    },
    {
      "epoch": 2.231808731808732,
      "grad_norm": 0.9033392071723938,
      "learning_rate": 6.153846153846155e-06,
      "loss": 0.0458,
      "step": 2147
    },
    {
      "epoch": 2.232848232848233,
      "grad_norm": 3.425045967102051,
      "learning_rate": 6.151536151536151e-06,
      "loss": 0.0722,
      "step": 2148
    },
    {
      "epoch": 2.233887733887734,
      "grad_norm": 0.7332718968391418,
      "learning_rate": 6.1492261492261496e-06,
      "loss": 0.0222,
      "step": 2149
    },
    {
      "epoch": 2.234927234927235,
      "grad_norm": 3.4322588443756104,
      "learning_rate": 6.146916146916148e-06,
      "loss": 0.144,
      "step": 2150
    },
    {
      "epoch": 2.235966735966736,
      "grad_norm": 0.6750631928443909,
      "learning_rate": 6.144606144606146e-06,
      "loss": 0.0257,
      "step": 2151
    },
    {
      "epoch": 2.237006237006237,
      "grad_norm": 10.790980339050293,
      "learning_rate": 6.1422961422961424e-06,
      "loss": 0.7807,
      "step": 2152
    },
    {
      "epoch": 2.238045738045738,
      "grad_norm": 1.1641652584075928,
      "learning_rate": 6.139986139986141e-06,
      "loss": 0.035,
      "step": 2153
    },
    {
      "epoch": 2.239085239085239,
      "grad_norm": 9.322601318359375,
      "learning_rate": 6.137676137676139e-06,
      "loss": 0.5508,
      "step": 2154
    },
    {
      "epoch": 2.24012474012474,
      "grad_norm": 1.2201451063156128,
      "learning_rate": 6.135366135366135e-06,
      "loss": 0.0281,
      "step": 2155
    },
    {
      "epoch": 2.241164241164241,
      "grad_norm": 0.011377022601664066,
      "learning_rate": 6.1330561330561335e-06,
      "loss": 0.0003,
      "step": 2156
    },
    {
      "epoch": 2.2422037422037424,
      "grad_norm": 0.3870171010494232,
      "learning_rate": 6.130746130746132e-06,
      "loss": 0.0147,
      "step": 2157
    },
    {
      "epoch": 2.2432432432432434,
      "grad_norm": 1.231650710105896,
      "learning_rate": 6.128436128436129e-06,
      "loss": 0.0324,
      "step": 2158
    },
    {
      "epoch": 2.2442827442827444,
      "grad_norm": 2.190561532974243,
      "learning_rate": 6.126126126126126e-06,
      "loss": 0.0829,
      "step": 2159
    },
    {
      "epoch": 2.2453222453222454,
      "grad_norm": 14.648651123046875,
      "learning_rate": 6.1238161238161246e-06,
      "loss": 0.7248,
      "step": 2160
    },
    {
      "epoch": 2.2463617463617465,
      "grad_norm": 0.22202709317207336,
      "learning_rate": 6.121506121506122e-06,
      "loss": 0.0044,
      "step": 2161
    },
    {
      "epoch": 2.2474012474012475,
      "grad_norm": 15.588959693908691,
      "learning_rate": 6.11919611919612e-06,
      "loss": 0.6235,
      "step": 2162
    },
    {
      "epoch": 2.2484407484407485,
      "grad_norm": 11.007787704467773,
      "learning_rate": 6.1168861168861175e-06,
      "loss": 0.5128,
      "step": 2163
    },
    {
      "epoch": 2.2494802494802495,
      "grad_norm": 0.2996562123298645,
      "learning_rate": 6.114576114576115e-06,
      "loss": 0.0055,
      "step": 2164
    },
    {
      "epoch": 2.2505197505197505,
      "grad_norm": 3.888937473297119,
      "learning_rate": 6.112266112266113e-06,
      "loss": 0.0838,
      "step": 2165
    },
    {
      "epoch": 2.2515592515592515,
      "grad_norm": 2.6077065467834473,
      "learning_rate": 6.10995610995611e-06,
      "loss": 0.0882,
      "step": 2166
    },
    {
      "epoch": 2.2525987525987525,
      "grad_norm": 0.4822016954421997,
      "learning_rate": 6.107646107646108e-06,
      "loss": 0.0107,
      "step": 2167
    },
    {
      "epoch": 2.2536382536382535,
      "grad_norm": 0.07543380558490753,
      "learning_rate": 6.105336105336106e-06,
      "loss": 0.0027,
      "step": 2168
    },
    {
      "epoch": 2.2546777546777546,
      "grad_norm": 0.3507520258426666,
      "learning_rate": 6.103026103026104e-06,
      "loss": 0.0069,
      "step": 2169
    },
    {
      "epoch": 2.2557172557172556,
      "grad_norm": 7.969033241271973,
      "learning_rate": 6.1007161007161006e-06,
      "loss": 0.9687,
      "step": 2170
    },
    {
      "epoch": 2.2567567567567566,
      "grad_norm": 6.0069050788879395,
      "learning_rate": 6.098406098406099e-06,
      "loss": 0.1818,
      "step": 2171
    },
    {
      "epoch": 2.257796257796258,
      "grad_norm": 13.834513664245605,
      "learning_rate": 6.096096096096097e-06,
      "loss": 0.8807,
      "step": 2172
    },
    {
      "epoch": 2.258835758835759,
      "grad_norm": 13.67642593383789,
      "learning_rate": 6.0937860937860934e-06,
      "loss": 0.6369,
      "step": 2173
    },
    {
      "epoch": 2.25987525987526,
      "grad_norm": 5.61348295211792,
      "learning_rate": 6.091476091476092e-06,
      "loss": 0.238,
      "step": 2174
    },
    {
      "epoch": 2.260914760914761,
      "grad_norm": 0.3030984699726105,
      "learning_rate": 6.08916608916609e-06,
      "loss": 0.0069,
      "step": 2175
    },
    {
      "epoch": 2.261954261954262,
      "grad_norm": 10.117189407348633,
      "learning_rate": 6.086856086856088e-06,
      "loss": 0.8853,
      "step": 2176
    },
    {
      "epoch": 2.262993762993763,
      "grad_norm": 9.245833396911621,
      "learning_rate": 6.0845460845460845e-06,
      "loss": 0.6308,
      "step": 2177
    },
    {
      "epoch": 2.264033264033264,
      "grad_norm": 0.23321442306041718,
      "learning_rate": 6.082236082236083e-06,
      "loss": 0.008,
      "step": 2178
    },
    {
      "epoch": 2.265072765072765,
      "grad_norm": 10.990395545959473,
      "learning_rate": 6.079926079926081e-06,
      "loss": 1.062,
      "step": 2179
    },
    {
      "epoch": 2.266112266112266,
      "grad_norm": 0.5124699473381042,
      "learning_rate": 6.077616077616078e-06,
      "loss": 0.0099,
      "step": 2180
    },
    {
      "epoch": 2.267151767151767,
      "grad_norm": 0.020312217995524406,
      "learning_rate": 6.0753060753060756e-06,
      "loss": 0.0003,
      "step": 2181
    },
    {
      "epoch": 2.268191268191268,
      "grad_norm": 3.825268507003784,
      "learning_rate": 6.072996072996074e-06,
      "loss": 0.1286,
      "step": 2182
    },
    {
      "epoch": 2.269230769230769,
      "grad_norm": 0.7043087482452393,
      "learning_rate": 6.070686070686071e-06,
      "loss": 0.0146,
      "step": 2183
    },
    {
      "epoch": 2.27027027027027,
      "grad_norm": 11.096062660217285,
      "learning_rate": 6.0683760683760684e-06,
      "loss": 1.0362,
      "step": 2184
    },
    {
      "epoch": 2.271309771309771,
      "grad_norm": 7.280477523803711,
      "learning_rate": 6.066066066066067e-06,
      "loss": 0.2911,
      "step": 2185
    },
    {
      "epoch": 2.272349272349272,
      "grad_norm": 0.15028230845928192,
      "learning_rate": 6.063756063756065e-06,
      "loss": 0.005,
      "step": 2186
    },
    {
      "epoch": 2.273388773388773,
      "grad_norm": 9.149582862854004,
      "learning_rate": 6.061446061446062e-06,
      "loss": 0.4077,
      "step": 2187
    },
    {
      "epoch": 2.274428274428274,
      "grad_norm": 2.9158825874328613,
      "learning_rate": 6.0591360591360595e-06,
      "loss": 0.0471,
      "step": 2188
    },
    {
      "epoch": 2.2754677754677752,
      "grad_norm": 0.09294558316469193,
      "learning_rate": 6.056826056826058e-06,
      "loss": 0.0015,
      "step": 2189
    },
    {
      "epoch": 2.2765072765072767,
      "grad_norm": 10.406388282775879,
      "learning_rate": 6.054516054516055e-06,
      "loss": 0.8003,
      "step": 2190
    },
    {
      "epoch": 2.2775467775467777,
      "grad_norm": 0.1650853008031845,
      "learning_rate": 6.052206052206052e-06,
      "loss": 0.0029,
      "step": 2191
    },
    {
      "epoch": 2.2785862785862787,
      "grad_norm": 7.553506374359131,
      "learning_rate": 6.049896049896051e-06,
      "loss": 0.3467,
      "step": 2192
    },
    {
      "epoch": 2.2796257796257797,
      "grad_norm": 0.027777332812547684,
      "learning_rate": 6.047586047586048e-06,
      "loss": 0.0005,
      "step": 2193
    },
    {
      "epoch": 2.2806652806652807,
      "grad_norm": 0.02463296428322792,
      "learning_rate": 6.045276045276046e-06,
      "loss": 0.0004,
      "step": 2194
    },
    {
      "epoch": 2.2817047817047817,
      "grad_norm": 15.498047828674316,
      "learning_rate": 6.0429660429660435e-06,
      "loss": 2.1882,
      "step": 2195
    },
    {
      "epoch": 2.2827442827442828,
      "grad_norm": 8.333736419677734,
      "learning_rate": 6.040656040656041e-06,
      "loss": 1.0823,
      "step": 2196
    },
    {
      "epoch": 2.2837837837837838,
      "grad_norm": 0.007465460803359747,
      "learning_rate": 6.038346038346039e-06,
      "loss": 0.0002,
      "step": 2197
    },
    {
      "epoch": 2.284823284823285,
      "grad_norm": 0.2807893753051758,
      "learning_rate": 6.036036036036037e-06,
      "loss": 0.0062,
      "step": 2198
    },
    {
      "epoch": 2.285862785862786,
      "grad_norm": 4.3804612159729,
      "learning_rate": 6.033726033726034e-06,
      "loss": 0.2496,
      "step": 2199
    },
    {
      "epoch": 2.286902286902287,
      "grad_norm": 1.8110418319702148,
      "learning_rate": 6.031416031416032e-06,
      "loss": 0.0364,
      "step": 2200
    },
    {
      "epoch": 2.287941787941788,
      "grad_norm": 1.0826025009155273,
      "learning_rate": 6.02910602910603e-06,
      "loss": 0.0204,
      "step": 2201
    },
    {
      "epoch": 2.288981288981289,
      "grad_norm": 0.04327649995684624,
      "learning_rate": 6.0267960267960266e-06,
      "loss": 0.0013,
      "step": 2202
    },
    {
      "epoch": 2.29002079002079,
      "grad_norm": 5.8153204917907715,
      "learning_rate": 6.024486024486025e-06,
      "loss": 0.2086,
      "step": 2203
    },
    {
      "epoch": 2.2910602910602913,
      "grad_norm": 0.011570545844733715,
      "learning_rate": 6.022176022176023e-06,
      "loss": 0.0003,
      "step": 2204
    },
    {
      "epoch": 2.2920997920997923,
      "grad_norm": 0.08150535076856613,
      "learning_rate": 6.019866019866021e-06,
      "loss": 0.0018,
      "step": 2205
    },
    {
      "epoch": 2.2931392931392933,
      "grad_norm": 8.084456443786621,
      "learning_rate": 6.017556017556018e-06,
      "loss": 0.7637,
      "step": 2206
    },
    {
      "epoch": 2.2941787941787943,
      "grad_norm": 0.06949927657842636,
      "learning_rate": 6.015246015246016e-06,
      "loss": 0.0016,
      "step": 2207
    },
    {
      "epoch": 2.2952182952182953,
      "grad_norm": 4.646961212158203,
      "learning_rate": 6.012936012936014e-06,
      "loss": 0.0951,
      "step": 2208
    },
    {
      "epoch": 2.2962577962577964,
      "grad_norm": 5.929029941558838,
      "learning_rate": 6.0106260106260105e-06,
      "loss": 0.3063,
      "step": 2209
    },
    {
      "epoch": 2.2972972972972974,
      "grad_norm": 2.86704683303833,
      "learning_rate": 6.008316008316009e-06,
      "loss": 0.0558,
      "step": 2210
    },
    {
      "epoch": 2.2983367983367984,
      "grad_norm": 8.642908096313477,
      "learning_rate": 6.006006006006007e-06,
      "loss": 0.4241,
      "step": 2211
    },
    {
      "epoch": 2.2993762993762994,
      "grad_norm": 3.363530397415161,
      "learning_rate": 6.003696003696004e-06,
      "loss": 0.0692,
      "step": 2212
    },
    {
      "epoch": 2.3004158004158004,
      "grad_norm": 8.524832725524902,
      "learning_rate": 6.0013860013860016e-06,
      "loss": 0.2695,
      "step": 2213
    },
    {
      "epoch": 2.3014553014553014,
      "grad_norm": 0.012578320689499378,
      "learning_rate": 5.999075999076e-06,
      "loss": 0.0002,
      "step": 2214
    },
    {
      "epoch": 2.3024948024948024,
      "grad_norm": 13.207572937011719,
      "learning_rate": 5.996765996765997e-06,
      "loss": 1.2314,
      "step": 2215
    },
    {
      "epoch": 2.3035343035343034,
      "grad_norm": 0.14138144254684448,
      "learning_rate": 5.994455994455995e-06,
      "loss": 0.0035,
      "step": 2216
    },
    {
      "epoch": 2.3045738045738045,
      "grad_norm": 0.029010187834501266,
      "learning_rate": 5.992145992145993e-06,
      "loss": 0.0006,
      "step": 2217
    },
    {
      "epoch": 2.3056133056133055,
      "grad_norm": 5.9447503089904785,
      "learning_rate": 5.98983598983599e-06,
      "loss": 0.2724,
      "step": 2218
    },
    {
      "epoch": 2.3066528066528065,
      "grad_norm": 9.484906196594238,
      "learning_rate": 5.987525987525988e-06,
      "loss": 0.7582,
      "step": 2219
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 9.741586685180664,
      "learning_rate": 5.9852159852159855e-06,
      "loss": 0.6975,
      "step": 2220
    },
    {
      "epoch": 2.3087318087318085,
      "grad_norm": 0.3729715347290039,
      "learning_rate": 5.982905982905983e-06,
      "loss": 0.0058,
      "step": 2221
    },
    {
      "epoch": 2.30977130977131,
      "grad_norm": 0.010007351636886597,
      "learning_rate": 5.980595980595981e-06,
      "loss": 0.0002,
      "step": 2222
    },
    {
      "epoch": 2.310810810810811,
      "grad_norm": 5.923950672149658,
      "learning_rate": 5.978285978285979e-06,
      "loss": 0.3793,
      "step": 2223
    },
    {
      "epoch": 2.311850311850312,
      "grad_norm": 6.51254415512085,
      "learning_rate": 5.975975975975976e-06,
      "loss": 0.5332,
      "step": 2224
    },
    {
      "epoch": 2.312889812889813,
      "grad_norm": 5.990417003631592,
      "learning_rate": 5.973665973665974e-06,
      "loss": 0.2406,
      "step": 2225
    },
    {
      "epoch": 2.313929313929314,
      "grad_norm": 2.215427875518799,
      "learning_rate": 5.971355971355972e-06,
      "loss": 0.0655,
      "step": 2226
    },
    {
      "epoch": 2.314968814968815,
      "grad_norm": 6.656970977783203,
      "learning_rate": 5.969045969045969e-06,
      "loss": 0.2158,
      "step": 2227
    },
    {
      "epoch": 2.316008316008316,
      "grad_norm": 11.145777702331543,
      "learning_rate": 5.966735966735967e-06,
      "loss": 0.3597,
      "step": 2228
    },
    {
      "epoch": 2.317047817047817,
      "grad_norm": 7.602084159851074,
      "learning_rate": 5.964425964425965e-06,
      "loss": 0.2979,
      "step": 2229
    },
    {
      "epoch": 2.318087318087318,
      "grad_norm": 12.412657737731934,
      "learning_rate": 5.962115962115963e-06,
      "loss": 0.7629,
      "step": 2230
    },
    {
      "epoch": 2.319126819126819,
      "grad_norm": 12.661900520324707,
      "learning_rate": 5.95980595980596e-06,
      "loss": 0.6984,
      "step": 2231
    },
    {
      "epoch": 2.32016632016632,
      "grad_norm": 5.584156513214111,
      "learning_rate": 5.957495957495958e-06,
      "loss": 0.1575,
      "step": 2232
    },
    {
      "epoch": 2.321205821205821,
      "grad_norm": 8.416341781616211,
      "learning_rate": 5.955185955185956e-06,
      "loss": 0.7292,
      "step": 2233
    },
    {
      "epoch": 2.322245322245322,
      "grad_norm": 16.33599090576172,
      "learning_rate": 5.952875952875954e-06,
      "loss": 1.921,
      "step": 2234
    },
    {
      "epoch": 2.323284823284823,
      "grad_norm": 8.614033699035645,
      "learning_rate": 5.950565950565951e-06,
      "loss": 0.269,
      "step": 2235
    },
    {
      "epoch": 2.3243243243243246,
      "grad_norm": 2.379479169845581,
      "learning_rate": 5.948255948255949e-06,
      "loss": 0.0411,
      "step": 2236
    },
    {
      "epoch": 2.3253638253638256,
      "grad_norm": 0.23624595999717712,
      "learning_rate": 5.945945945945947e-06,
      "loss": 0.0092,
      "step": 2237
    },
    {
      "epoch": 2.3264033264033266,
      "grad_norm": 2.8122398853302,
      "learning_rate": 5.943635943635944e-06,
      "loss": 0.1048,
      "step": 2238
    },
    {
      "epoch": 2.3274428274428276,
      "grad_norm": 1.7436000108718872,
      "learning_rate": 5.941325941325942e-06,
      "loss": 0.2694,
      "step": 2239
    },
    {
      "epoch": 2.3284823284823286,
      "grad_norm": 1.277492642402649,
      "learning_rate": 5.93901593901594e-06,
      "loss": 0.023,
      "step": 2240
    },
    {
      "epoch": 2.3295218295218296,
      "grad_norm": 3.3322794437408447,
      "learning_rate": 5.936705936705937e-06,
      "loss": 0.274,
      "step": 2241
    },
    {
      "epoch": 2.3305613305613306,
      "grad_norm": 5.62230110168457,
      "learning_rate": 5.934395934395935e-06,
      "loss": 0.2909,
      "step": 2242
    },
    {
      "epoch": 2.3316008316008316,
      "grad_norm": 17.122243881225586,
      "learning_rate": 5.932085932085933e-06,
      "loss": 0.2684,
      "step": 2243
    },
    {
      "epoch": 2.3326403326403327,
      "grad_norm": 0.5729025602340698,
      "learning_rate": 5.92977592977593e-06,
      "loss": 0.0081,
      "step": 2244
    },
    {
      "epoch": 2.3336798336798337,
      "grad_norm": 0.029930051416158676,
      "learning_rate": 5.927465927465928e-06,
      "loss": 0.0007,
      "step": 2245
    },
    {
      "epoch": 2.3347193347193347,
      "grad_norm": 0.48070666193962097,
      "learning_rate": 5.925155925155926e-06,
      "loss": 0.0102,
      "step": 2246
    },
    {
      "epoch": 2.3357588357588357,
      "grad_norm": 4.338085651397705,
      "learning_rate": 5.922845922845923e-06,
      "loss": 0.0833,
      "step": 2247
    },
    {
      "epoch": 2.3367983367983367,
      "grad_norm": 2.9266693592071533,
      "learning_rate": 5.920535920535921e-06,
      "loss": 0.2003,
      "step": 2248
    },
    {
      "epoch": 2.3378378378378377,
      "grad_norm": 10.988312721252441,
      "learning_rate": 5.918225918225919e-06,
      "loss": 1.367,
      "step": 2249
    },
    {
      "epoch": 2.3388773388773387,
      "grad_norm": 2.4815878868103027,
      "learning_rate": 5.915915915915916e-06,
      "loss": 0.137,
      "step": 2250
    },
    {
      "epoch": 2.3399168399168397,
      "grad_norm": 7.028376579284668,
      "learning_rate": 5.913605913605914e-06,
      "loss": 0.3135,
      "step": 2251
    },
    {
      "epoch": 2.3409563409563408,
      "grad_norm": 9.232767105102539,
      "learning_rate": 5.911295911295912e-06,
      "loss": 0.752,
      "step": 2252
    },
    {
      "epoch": 2.3419958419958418,
      "grad_norm": 9.30560302734375,
      "learning_rate": 5.908985908985909e-06,
      "loss": 0.8332,
      "step": 2253
    },
    {
      "epoch": 2.343035343035343,
      "grad_norm": 1.2369141578674316,
      "learning_rate": 5.906675906675907e-06,
      "loss": 0.0361,
      "step": 2254
    },
    {
      "epoch": 2.3440748440748442,
      "grad_norm": 0.7799108028411865,
      "learning_rate": 5.904365904365905e-06,
      "loss": 0.0149,
      "step": 2255
    },
    {
      "epoch": 2.3451143451143452,
      "grad_norm": 1.8956689834594727,
      "learning_rate": 5.902055902055902e-06,
      "loss": 0.0897,
      "step": 2256
    },
    {
      "epoch": 2.3461538461538463,
      "grad_norm": 0.285441517829895,
      "learning_rate": 5.8997458997459e-06,
      "loss": 0.005,
      "step": 2257
    },
    {
      "epoch": 2.3471933471933473,
      "grad_norm": 0.18665236234664917,
      "learning_rate": 5.897435897435898e-06,
      "loss": 0.0059,
      "step": 2258
    },
    {
      "epoch": 2.3482328482328483,
      "grad_norm": 0.08902893960475922,
      "learning_rate": 5.895125895125896e-06,
      "loss": 0.0013,
      "step": 2259
    },
    {
      "epoch": 2.3492723492723493,
      "grad_norm": 12.768613815307617,
      "learning_rate": 5.892815892815893e-06,
      "loss": 1.074,
      "step": 2260
    },
    {
      "epoch": 2.3503118503118503,
      "grad_norm": 4.8039774894714355,
      "learning_rate": 5.890505890505891e-06,
      "loss": 0.1545,
      "step": 2261
    },
    {
      "epoch": 2.3513513513513513,
      "grad_norm": 0.005349677987396717,
      "learning_rate": 5.888195888195889e-06,
      "loss": 0.0001,
      "step": 2262
    },
    {
      "epoch": 2.3523908523908523,
      "grad_norm": 8.962315559387207,
      "learning_rate": 5.885885885885886e-06,
      "loss": 0.4207,
      "step": 2263
    },
    {
      "epoch": 2.3534303534303533,
      "grad_norm": 12.141010284423828,
      "learning_rate": 5.883575883575884e-06,
      "loss": 0.9134,
      "step": 2264
    },
    {
      "epoch": 2.3544698544698544,
      "grad_norm": 1.6301746368408203,
      "learning_rate": 5.881265881265882e-06,
      "loss": 0.0752,
      "step": 2265
    },
    {
      "epoch": 2.3555093555093554,
      "grad_norm": 9.701839447021484,
      "learning_rate": 5.8789558789558794e-06,
      "loss": 0.329,
      "step": 2266
    },
    {
      "epoch": 2.3565488565488564,
      "grad_norm": 0.2083820402622223,
      "learning_rate": 5.876645876645877e-06,
      "loss": 0.0077,
      "step": 2267
    },
    {
      "epoch": 2.357588357588358,
      "grad_norm": 0.5354583263397217,
      "learning_rate": 5.874335874335875e-06,
      "loss": 0.011,
      "step": 2268
    },
    {
      "epoch": 2.358627858627859,
      "grad_norm": 0.549380898475647,
      "learning_rate": 5.872025872025872e-06,
      "loss": 0.0078,
      "step": 2269
    },
    {
      "epoch": 2.35966735966736,
      "grad_norm": 0.06869293004274368,
      "learning_rate": 5.8697158697158705e-06,
      "loss": 0.0012,
      "step": 2270
    },
    {
      "epoch": 2.360706860706861,
      "grad_norm": 4.068930149078369,
      "learning_rate": 5.867405867405868e-06,
      "loss": 0.1136,
      "step": 2271
    },
    {
      "epoch": 2.361746361746362,
      "grad_norm": 1.1226253509521484,
      "learning_rate": 5.865095865095865e-06,
      "loss": 0.0287,
      "step": 2272
    },
    {
      "epoch": 2.362785862785863,
      "grad_norm": 9.672595977783203,
      "learning_rate": 5.862785862785863e-06,
      "loss": 0.152,
      "step": 2273
    },
    {
      "epoch": 2.363825363825364,
      "grad_norm": 6.1760077476501465,
      "learning_rate": 5.860475860475861e-06,
      "loss": 0.3447,
      "step": 2274
    },
    {
      "epoch": 2.364864864864865,
      "grad_norm": 5.4050421714782715,
      "learning_rate": 5.858165858165858e-06,
      "loss": 0.0732,
      "step": 2275
    },
    {
      "epoch": 2.365904365904366,
      "grad_norm": 1.3560022115707397,
      "learning_rate": 5.855855855855856e-06,
      "loss": 0.0473,
      "step": 2276
    },
    {
      "epoch": 2.366943866943867,
      "grad_norm": 5.8554887771606445,
      "learning_rate": 5.8535458535458544e-06,
      "loss": 0.1202,
      "step": 2277
    },
    {
      "epoch": 2.367983367983368,
      "grad_norm": 11.231568336486816,
      "learning_rate": 5.851235851235851e-06,
      "loss": 1.1266,
      "step": 2278
    },
    {
      "epoch": 2.369022869022869,
      "grad_norm": 0.5555443167686462,
      "learning_rate": 5.848925848925849e-06,
      "loss": 0.0119,
      "step": 2279
    },
    {
      "epoch": 2.37006237006237,
      "grad_norm": 2.8198113441467285,
      "learning_rate": 5.846615846615847e-06,
      "loss": 0.175,
      "step": 2280
    },
    {
      "epoch": 2.371101871101871,
      "grad_norm": 1.1452579498291016,
      "learning_rate": 5.844305844305845e-06,
      "loss": 0.0255,
      "step": 2281
    },
    {
      "epoch": 2.372141372141372,
      "grad_norm": 3.1059226989746094,
      "learning_rate": 5.841995841995842e-06,
      "loss": 0.0457,
      "step": 2282
    },
    {
      "epoch": 2.373180873180873,
      "grad_norm": 1.5410337448120117,
      "learning_rate": 5.83968583968584e-06,
      "loss": 0.0515,
      "step": 2283
    },
    {
      "epoch": 2.374220374220374,
      "grad_norm": 4.821375846862793,
      "learning_rate": 5.837375837375838e-06,
      "loss": 0.1617,
      "step": 2284
    },
    {
      "epoch": 2.375259875259875,
      "grad_norm": 6.064678192138672,
      "learning_rate": 5.835065835065835e-06,
      "loss": 0.512,
      "step": 2285
    },
    {
      "epoch": 2.3762993762993765,
      "grad_norm": 1.9220199584960938,
      "learning_rate": 5.832755832755833e-06,
      "loss": 0.0384,
      "step": 2286
    },
    {
      "epoch": 2.3773388773388775,
      "grad_norm": 4.801299095153809,
      "learning_rate": 5.830445830445831e-06,
      "loss": 0.2309,
      "step": 2287
    },
    {
      "epoch": 2.3783783783783785,
      "grad_norm": 3.1393089294433594,
      "learning_rate": 5.8281358281358295e-06,
      "loss": 0.0715,
      "step": 2288
    },
    {
      "epoch": 2.3794178794178795,
      "grad_norm": 3.297896146774292,
      "learning_rate": 5.825825825825826e-06,
      "loss": 0.2937,
      "step": 2289
    },
    {
      "epoch": 2.3804573804573805,
      "grad_norm": 0.050158146768808365,
      "learning_rate": 5.823515823515824e-06,
      "loss": 0.001,
      "step": 2290
    },
    {
      "epoch": 2.3814968814968815,
      "grad_norm": 11.856502532958984,
      "learning_rate": 5.821205821205822e-06,
      "loss": 0.5053,
      "step": 2291
    },
    {
      "epoch": 2.3825363825363826,
      "grad_norm": 7.975099086761475,
      "learning_rate": 5.818895818895819e-06,
      "loss": 0.2572,
      "step": 2292
    },
    {
      "epoch": 2.3835758835758836,
      "grad_norm": 11.382410049438477,
      "learning_rate": 5.816585816585817e-06,
      "loss": 0.6613,
      "step": 2293
    },
    {
      "epoch": 2.3846153846153846,
      "grad_norm": 2.8917577266693115,
      "learning_rate": 5.814275814275815e-06,
      "loss": 0.1031,
      "step": 2294
    },
    {
      "epoch": 2.3856548856548856,
      "grad_norm": 4.3138298988342285,
      "learning_rate": 5.8119658119658126e-06,
      "loss": 0.1378,
      "step": 2295
    },
    {
      "epoch": 2.3866943866943866,
      "grad_norm": 6.972907066345215,
      "learning_rate": 5.80965580965581e-06,
      "loss": 0.2819,
      "step": 2296
    },
    {
      "epoch": 2.3877338877338876,
      "grad_norm": 3.7977538108825684,
      "learning_rate": 5.807345807345808e-06,
      "loss": 0.2146,
      "step": 2297
    },
    {
      "epoch": 2.3887733887733886,
      "grad_norm": 0.20148056745529175,
      "learning_rate": 5.8050358050358054e-06,
      "loss": 0.003,
      "step": 2298
    },
    {
      "epoch": 2.3898128898128896,
      "grad_norm": 9.711847305297852,
      "learning_rate": 5.802725802725803e-06,
      "loss": 1.2043,
      "step": 2299
    },
    {
      "epoch": 2.390852390852391,
      "grad_norm": 0.4300665855407715,
      "learning_rate": 5.800415800415801e-06,
      "loss": 0.0093,
      "step": 2300
    },
    {
      "epoch": 2.391891891891892,
      "grad_norm": 4.982692718505859,
      "learning_rate": 5.798105798105798e-06,
      "loss": 0.1541,
      "step": 2301
    },
    {
      "epoch": 2.392931392931393,
      "grad_norm": 13.709635734558105,
      "learning_rate": 5.7957957957957965e-06,
      "loss": 0.8311,
      "step": 2302
    },
    {
      "epoch": 2.393970893970894,
      "grad_norm": 8.99238109588623,
      "learning_rate": 5.793485793485794e-06,
      "loss": 0.3719,
      "step": 2303
    },
    {
      "epoch": 2.395010395010395,
      "grad_norm": 0.019779542461037636,
      "learning_rate": 5.791175791175791e-06,
      "loss": 0.0005,
      "step": 2304
    },
    {
      "epoch": 2.396049896049896,
      "grad_norm": 1.7408281564712524,
      "learning_rate": 5.788865788865789e-06,
      "loss": 0.0302,
      "step": 2305
    },
    {
      "epoch": 2.397089397089397,
      "grad_norm": 0.5736076831817627,
      "learning_rate": 5.7865557865557876e-06,
      "loss": 0.0094,
      "step": 2306
    },
    {
      "epoch": 2.398128898128898,
      "grad_norm": 0.351347953081131,
      "learning_rate": 5.784245784245784e-06,
      "loss": 0.0099,
      "step": 2307
    },
    {
      "epoch": 2.399168399168399,
      "grad_norm": 1.1329315900802612,
      "learning_rate": 5.781935781935782e-06,
      "loss": 0.0239,
      "step": 2308
    },
    {
      "epoch": 2.4002079002079,
      "grad_norm": 0.43927931785583496,
      "learning_rate": 5.7796257796257805e-06,
      "loss": 0.0071,
      "step": 2309
    },
    {
      "epoch": 2.401247401247401,
      "grad_norm": 9.422633171081543,
      "learning_rate": 5.777315777315777e-06,
      "loss": 0.2959,
      "step": 2310
    },
    {
      "epoch": 2.4022869022869022,
      "grad_norm": 22.3996639251709,
      "learning_rate": 5.775005775005775e-06,
      "loss": 1.3917,
      "step": 2311
    },
    {
      "epoch": 2.4033264033264032,
      "grad_norm": 7.138119697570801,
      "learning_rate": 5.772695772695773e-06,
      "loss": 0.2916,
      "step": 2312
    },
    {
      "epoch": 2.4043659043659042,
      "grad_norm": 0.04603596776723862,
      "learning_rate": 5.7703857703857715e-06,
      "loss": 0.0008,
      "step": 2313
    },
    {
      "epoch": 2.4054054054054053,
      "grad_norm": 0.22304362058639526,
      "learning_rate": 5.768075768075768e-06,
      "loss": 0.0032,
      "step": 2314
    },
    {
      "epoch": 2.4064449064449063,
      "grad_norm": 0.44586488604545593,
      "learning_rate": 5.765765765765766e-06,
      "loss": 0.0142,
      "step": 2315
    },
    {
      "epoch": 2.4074844074844073,
      "grad_norm": 1.6813427209854126,
      "learning_rate": 5.763455763455764e-06,
      "loss": 0.0507,
      "step": 2316
    },
    {
      "epoch": 2.4085239085239083,
      "grad_norm": 0.3971876800060272,
      "learning_rate": 5.761145761145761e-06,
      "loss": 0.0072,
      "step": 2317
    },
    {
      "epoch": 2.4095634095634098,
      "grad_norm": 0.08452802896499634,
      "learning_rate": 5.758835758835759e-06,
      "loss": 0.002,
      "step": 2318
    },
    {
      "epoch": 2.4106029106029108,
      "grad_norm": 11.850491523742676,
      "learning_rate": 5.756525756525757e-06,
      "loss": 0.4145,
      "step": 2319
    },
    {
      "epoch": 2.4116424116424118,
      "grad_norm": 2.9328062534332275,
      "learning_rate": 5.7542157542157555e-06,
      "loss": 0.0477,
      "step": 2320
    },
    {
      "epoch": 2.412681912681913,
      "grad_norm": 3.1229493618011475,
      "learning_rate": 5.751905751905752e-06,
      "loss": 0.0631,
      "step": 2321
    },
    {
      "epoch": 2.413721413721414,
      "grad_norm": 0.15812994539737701,
      "learning_rate": 5.74959574959575e-06,
      "loss": 0.005,
      "step": 2322
    },
    {
      "epoch": 2.414760914760915,
      "grad_norm": 0.660758376121521,
      "learning_rate": 5.747285747285748e-06,
      "loss": 0.0081,
      "step": 2323
    },
    {
      "epoch": 2.415800415800416,
      "grad_norm": 16.216384887695312,
      "learning_rate": 5.744975744975746e-06,
      "loss": 0.7467,
      "step": 2324
    },
    {
      "epoch": 2.416839916839917,
      "grad_norm": 8.676569938659668,
      "learning_rate": 5.742665742665743e-06,
      "loss": 0.3671,
      "step": 2325
    },
    {
      "epoch": 2.417879417879418,
      "grad_norm": 2.7714102268218994,
      "learning_rate": 5.740355740355741e-06,
      "loss": 0.1326,
      "step": 2326
    },
    {
      "epoch": 2.418918918918919,
      "grad_norm": 4.570979118347168,
      "learning_rate": 5.7380457380457386e-06,
      "loss": 0.1253,
      "step": 2327
    },
    {
      "epoch": 2.41995841995842,
      "grad_norm": 9.01137924194336,
      "learning_rate": 5.735735735735736e-06,
      "loss": 0.7805,
      "step": 2328
    },
    {
      "epoch": 2.420997920997921,
      "grad_norm": 10.329658508300781,
      "learning_rate": 5.733425733425734e-06,
      "loss": 0.3824,
      "step": 2329
    },
    {
      "epoch": 2.422037422037422,
      "grad_norm": 0.7224492430686951,
      "learning_rate": 5.7311157311157314e-06,
      "loss": 0.0235,
      "step": 2330
    },
    {
      "epoch": 2.423076923076923,
      "grad_norm": 0.059857260435819626,
      "learning_rate": 5.72880572880573e-06,
      "loss": 0.0011,
      "step": 2331
    },
    {
      "epoch": 2.4241164241164244,
      "grad_norm": 0.3503856956958771,
      "learning_rate": 5.726495726495727e-06,
      "loss": 0.0066,
      "step": 2332
    },
    {
      "epoch": 2.4251559251559254,
      "grad_norm": 0.03683709353208542,
      "learning_rate": 5.724185724185724e-06,
      "loss": 0.0008,
      "step": 2333
    },
    {
      "epoch": 2.4261954261954264,
      "grad_norm": 0.03195773437619209,
      "learning_rate": 5.7218757218757225e-06,
      "loss": 0.0008,
      "step": 2334
    },
    {
      "epoch": 2.4272349272349274,
      "grad_norm": 2.0734236240386963,
      "learning_rate": 5.71956571956572e-06,
      "loss": 0.0504,
      "step": 2335
    },
    {
      "epoch": 2.4282744282744284,
      "grad_norm": 1.467271089553833,
      "learning_rate": 5.717255717255717e-06,
      "loss": 0.0275,
      "step": 2336
    },
    {
      "epoch": 2.4293139293139294,
      "grad_norm": 10.107986450195312,
      "learning_rate": 5.714945714945715e-06,
      "loss": 0.3636,
      "step": 2337
    },
    {
      "epoch": 2.4303534303534304,
      "grad_norm": 4.867897033691406,
      "learning_rate": 5.712635712635714e-06,
      "loss": 0.1089,
      "step": 2338
    },
    {
      "epoch": 2.4313929313929314,
      "grad_norm": 1.6725810766220093,
      "learning_rate": 5.71032571032571e-06,
      "loss": 0.0591,
      "step": 2339
    },
    {
      "epoch": 2.4324324324324325,
      "grad_norm": 5.271655082702637,
      "learning_rate": 5.708015708015708e-06,
      "loss": 0.1665,
      "step": 2340
    },
    {
      "epoch": 2.4334719334719335,
      "grad_norm": 0.596553385257721,
      "learning_rate": 5.7057057057057065e-06,
      "loss": 0.0163,
      "step": 2341
    },
    {
      "epoch": 2.4345114345114345,
      "grad_norm": 8.481026649475098,
      "learning_rate": 5.703395703395705e-06,
      "loss": 0.351,
      "step": 2342
    },
    {
      "epoch": 2.4355509355509355,
      "grad_norm": 0.14047005772590637,
      "learning_rate": 5.701085701085701e-06,
      "loss": 0.003,
      "step": 2343
    },
    {
      "epoch": 2.4365904365904365,
      "grad_norm": 1.8173028230667114,
      "learning_rate": 5.698775698775699e-06,
      "loss": 0.036,
      "step": 2344
    },
    {
      "epoch": 2.4376299376299375,
      "grad_norm": 0.9283024668693542,
      "learning_rate": 5.6964656964656975e-06,
      "loss": 0.0173,
      "step": 2345
    },
    {
      "epoch": 2.4386694386694385,
      "grad_norm": 10.197988510131836,
      "learning_rate": 5.694155694155694e-06,
      "loss": 0.4031,
      "step": 2346
    },
    {
      "epoch": 2.4397089397089395,
      "grad_norm": 9.023080825805664,
      "learning_rate": 5.691845691845692e-06,
      "loss": 0.8144,
      "step": 2347
    },
    {
      "epoch": 2.4407484407484406,
      "grad_norm": 0.4014405906200409,
      "learning_rate": 5.68953568953569e-06,
      "loss": 0.008,
      "step": 2348
    },
    {
      "epoch": 2.4417879417879416,
      "grad_norm": 1.333750605583191,
      "learning_rate": 5.687225687225688e-06,
      "loss": 0.046,
      "step": 2349
    },
    {
      "epoch": 2.442827442827443,
      "grad_norm": 4.866448879241943,
      "learning_rate": 5.684915684915685e-06,
      "loss": 0.0696,
      "step": 2350
    },
    {
      "epoch": 2.443866943866944,
      "grad_norm": 4.256906509399414,
      "learning_rate": 5.682605682605683e-06,
      "loss": 0.1184,
      "step": 2351
    },
    {
      "epoch": 2.444906444906445,
      "grad_norm": 6.979046821594238,
      "learning_rate": 5.680295680295681e-06,
      "loss": 0.1771,
      "step": 2352
    },
    {
      "epoch": 2.445945945945946,
      "grad_norm": 6.416190147399902,
      "learning_rate": 5.677985677985678e-06,
      "loss": 0.4255,
      "step": 2353
    },
    {
      "epoch": 2.446985446985447,
      "grad_norm": 0.5552463531494141,
      "learning_rate": 5.675675675675676e-06,
      "loss": 0.0138,
      "step": 2354
    },
    {
      "epoch": 2.448024948024948,
      "grad_norm": 3.588365077972412,
      "learning_rate": 5.6733656733656735e-06,
      "loss": 0.1254,
      "step": 2355
    },
    {
      "epoch": 2.449064449064449,
      "grad_norm": 0.11385034769773483,
      "learning_rate": 5.671055671055672e-06,
      "loss": 0.0024,
      "step": 2356
    },
    {
      "epoch": 2.45010395010395,
      "grad_norm": 3.5646817684173584,
      "learning_rate": 5.668745668745669e-06,
      "loss": 0.0715,
      "step": 2357
    },
    {
      "epoch": 2.451143451143451,
      "grad_norm": 2.0259387493133545,
      "learning_rate": 5.666435666435666e-06,
      "loss": 0.042,
      "step": 2358
    },
    {
      "epoch": 2.452182952182952,
      "grad_norm": 3.4722325801849365,
      "learning_rate": 5.6641256641256646e-06,
      "loss": 0.086,
      "step": 2359
    },
    {
      "epoch": 2.453222453222453,
      "grad_norm": 1.5582636594772339,
      "learning_rate": 5.661815661815663e-06,
      "loss": 0.0273,
      "step": 2360
    },
    {
      "epoch": 2.454261954261954,
      "grad_norm": 3.767815589904785,
      "learning_rate": 5.659505659505659e-06,
      "loss": 0.0896,
      "step": 2361
    },
    {
      "epoch": 2.455301455301455,
      "grad_norm": 4.259250640869141,
      "learning_rate": 5.6571956571956575e-06,
      "loss": 0.1208,
      "step": 2362
    },
    {
      "epoch": 2.456340956340956,
      "grad_norm": 0.0018083971226587892,
      "learning_rate": 5.654885654885656e-06,
      "loss": 0.0,
      "step": 2363
    },
    {
      "epoch": 2.4573804573804576,
      "grad_norm": 2.261906385421753,
      "learning_rate": 5.652575652575652e-06,
      "loss": 0.0576,
      "step": 2364
    },
    {
      "epoch": 2.4584199584199586,
      "grad_norm": 0.015318424440920353,
      "learning_rate": 5.65026565026565e-06,
      "loss": 0.0005,
      "step": 2365
    },
    {
      "epoch": 2.4594594594594597,
      "grad_norm": 1.0688711404800415,
      "learning_rate": 5.6479556479556485e-06,
      "loss": 0.0209,
      "step": 2366
    },
    {
      "epoch": 2.4604989604989607,
      "grad_norm": 9.626909255981445,
      "learning_rate": 5.645645645645647e-06,
      "loss": 1.2241,
      "step": 2367
    },
    {
      "epoch": 2.4615384615384617,
      "grad_norm": 0.9243369102478027,
      "learning_rate": 5.643335643335643e-06,
      "loss": 0.0235,
      "step": 2368
    },
    {
      "epoch": 2.4625779625779627,
      "grad_norm": 4.78314208984375,
      "learning_rate": 5.641025641025641e-06,
      "loss": 0.2238,
      "step": 2369
    },
    {
      "epoch": 2.4636174636174637,
      "grad_norm": 0.6715689301490784,
      "learning_rate": 5.63871563871564e-06,
      "loss": 0.0188,
      "step": 2370
    },
    {
      "epoch": 2.4646569646569647,
      "grad_norm": 0.03074975498020649,
      "learning_rate": 5.636405636405636e-06,
      "loss": 0.0007,
      "step": 2371
    },
    {
      "epoch": 2.4656964656964657,
      "grad_norm": 11.54360294342041,
      "learning_rate": 5.634095634095634e-06,
      "loss": 0.8552,
      "step": 2372
    },
    {
      "epoch": 2.4667359667359667,
      "grad_norm": 0.5170840620994568,
      "learning_rate": 5.6317856317856325e-06,
      "loss": 0.0108,
      "step": 2373
    },
    {
      "epoch": 2.4677754677754677,
      "grad_norm": 5.690970420837402,
      "learning_rate": 5.629475629475631e-06,
      "loss": 0.18,
      "step": 2374
    },
    {
      "epoch": 2.4688149688149688,
      "grad_norm": 10.498955726623535,
      "learning_rate": 5.627165627165627e-06,
      "loss": 1.3798,
      "step": 2375
    },
    {
      "epoch": 2.4698544698544698,
      "grad_norm": 7.402065753936768,
      "learning_rate": 5.624855624855625e-06,
      "loss": 0.2798,
      "step": 2376
    },
    {
      "epoch": 2.470893970893971,
      "grad_norm": 2.759883403778076,
      "learning_rate": 5.6225456225456235e-06,
      "loss": 0.1178,
      "step": 2377
    },
    {
      "epoch": 2.471933471933472,
      "grad_norm": 0.09848170727491379,
      "learning_rate": 5.620235620235621e-06,
      "loss": 0.0019,
      "step": 2378
    },
    {
      "epoch": 2.472972972972973,
      "grad_norm": 8.388592720031738,
      "learning_rate": 5.617925617925618e-06,
      "loss": 0.2469,
      "step": 2379
    },
    {
      "epoch": 2.474012474012474,
      "grad_norm": 5.854394912719727,
      "learning_rate": 5.615615615615616e-06,
      "loss": 0.3582,
      "step": 2380
    },
    {
      "epoch": 2.475051975051975,
      "grad_norm": 7.211421966552734,
      "learning_rate": 5.613305613305614e-06,
      "loss": 0.6539,
      "step": 2381
    },
    {
      "epoch": 2.4760914760914763,
      "grad_norm": 0.45193174481391907,
      "learning_rate": 5.610995610995611e-06,
      "loss": 0.0105,
      "step": 2382
    },
    {
      "epoch": 2.4771309771309773,
      "grad_norm": 0.1756490170955658,
      "learning_rate": 5.608685608685609e-06,
      "loss": 0.0051,
      "step": 2383
    },
    {
      "epoch": 2.4781704781704783,
      "grad_norm": 8.903878211975098,
      "learning_rate": 5.606375606375607e-06,
      "loss": 0.363,
      "step": 2384
    },
    {
      "epoch": 2.4792099792099793,
      "grad_norm": 4.330974102020264,
      "learning_rate": 5.604065604065605e-06,
      "loss": 0.1238,
      "step": 2385
    },
    {
      "epoch": 2.4802494802494803,
      "grad_norm": 8.367501258850098,
      "learning_rate": 5.601755601755602e-06,
      "loss": 0.3349,
      "step": 2386
    },
    {
      "epoch": 2.4812889812889813,
      "grad_norm": 0.2912248373031616,
      "learning_rate": 5.5994455994455995e-06,
      "loss": 0.0045,
      "step": 2387
    },
    {
      "epoch": 2.4823284823284824,
      "grad_norm": 3.1457712650299072,
      "learning_rate": 5.597135597135598e-06,
      "loss": 0.0796,
      "step": 2388
    },
    {
      "epoch": 2.4833679833679834,
      "grad_norm": 5.04201078414917,
      "learning_rate": 5.594825594825595e-06,
      "loss": 0.2289,
      "step": 2389
    },
    {
      "epoch": 2.4844074844074844,
      "grad_norm": 0.1276787668466568,
      "learning_rate": 5.592515592515592e-06,
      "loss": 0.0032,
      "step": 2390
    },
    {
      "epoch": 2.4854469854469854,
      "grad_norm": 12.047901153564453,
      "learning_rate": 5.590205590205591e-06,
      "loss": 0.4329,
      "step": 2391
    },
    {
      "epoch": 2.4864864864864864,
      "grad_norm": 9.832515716552734,
      "learning_rate": 5.587895587895589e-06,
      "loss": 0.9411,
      "step": 2392
    },
    {
      "epoch": 2.4875259875259874,
      "grad_norm": 10.79465103149414,
      "learning_rate": 5.585585585585585e-06,
      "loss": 0.2082,
      "step": 2393
    },
    {
      "epoch": 2.4885654885654884,
      "grad_norm": 2.7653582096099854,
      "learning_rate": 5.5832755832755835e-06,
      "loss": 0.0389,
      "step": 2394
    },
    {
      "epoch": 2.4896049896049894,
      "grad_norm": 7.273566722869873,
      "learning_rate": 5.580965580965582e-06,
      "loss": 0.1308,
      "step": 2395
    },
    {
      "epoch": 2.490644490644491,
      "grad_norm": 0.6354008316993713,
      "learning_rate": 5.57865557865558e-06,
      "loss": 0.0125,
      "step": 2396
    },
    {
      "epoch": 2.491683991683992,
      "grad_norm": 0.014644395560026169,
      "learning_rate": 5.576345576345576e-06,
      "loss": 0.0004,
      "step": 2397
    },
    {
      "epoch": 2.492723492723493,
      "grad_norm": 8.341543197631836,
      "learning_rate": 5.5740355740355745e-06,
      "loss": 1.623,
      "step": 2398
    },
    {
      "epoch": 2.493762993762994,
      "grad_norm": 5.564713001251221,
      "learning_rate": 5.571725571725573e-06,
      "loss": 0.0785,
      "step": 2399
    },
    {
      "epoch": 2.494802494802495,
      "grad_norm": 0.19482921063899994,
      "learning_rate": 5.569415569415569e-06,
      "loss": 0.004,
      "step": 2400
    },
    {
      "epoch": 2.495841995841996,
      "grad_norm": 2.786550998687744,
      "learning_rate": 5.567105567105567e-06,
      "loss": 0.0618,
      "step": 2401
    },
    {
      "epoch": 2.496881496881497,
      "grad_norm": 10.698128700256348,
      "learning_rate": 5.564795564795566e-06,
      "loss": 0.4151,
      "step": 2402
    },
    {
      "epoch": 2.497920997920998,
      "grad_norm": 8.118881225585938,
      "learning_rate": 5.562485562485564e-06,
      "loss": 0.297,
      "step": 2403
    },
    {
      "epoch": 2.498960498960499,
      "grad_norm": 0.05869365483522415,
      "learning_rate": 5.56017556017556e-06,
      "loss": 0.0013,
      "step": 2404
    },
    {
      "epoch": 2.5,
      "grad_norm": 6.432159423828125,
      "learning_rate": 5.5578655578655585e-06,
      "loss": 0.1399,
      "step": 2405
    },
    {
      "epoch": 2.501039501039501,
      "grad_norm": 7.843135356903076,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.4493,
      "step": 2406
    },
    {
      "epoch": 2.502079002079002,
      "grad_norm": 13.558523178100586,
      "learning_rate": 5.553245553245554e-06,
      "loss": 0.1005,
      "step": 2407
    },
    {
      "epoch": 2.503118503118503,
      "grad_norm": 0.7968465685844421,
      "learning_rate": 5.550935550935551e-06,
      "loss": 0.0185,
      "step": 2408
    },
    {
      "epoch": 2.504158004158004,
      "grad_norm": 2.6324613094329834,
      "learning_rate": 5.5486255486255495e-06,
      "loss": 0.0922,
      "step": 2409
    },
    {
      "epoch": 2.505197505197505,
      "grad_norm": 0.7506412267684937,
      "learning_rate": 5.546315546315547e-06,
      "loss": 0.0148,
      "step": 2410
    },
    {
      "epoch": 2.506237006237006,
      "grad_norm": 0.028840284794569016,
      "learning_rate": 5.544005544005544e-06,
      "loss": 0.001,
      "step": 2411
    },
    {
      "epoch": 2.507276507276507,
      "grad_norm": 6.280975341796875,
      "learning_rate": 5.5416955416955424e-06,
      "loss": 0.2602,
      "step": 2412
    },
    {
      "epoch": 2.508316008316008,
      "grad_norm": 2.0573740005493164,
      "learning_rate": 5.53938553938554e-06,
      "loss": 0.0502,
      "step": 2413
    },
    {
      "epoch": 2.509355509355509,
      "grad_norm": 9.995739936828613,
      "learning_rate": 5.537075537075538e-06,
      "loss": 0.9421,
      "step": 2414
    },
    {
      "epoch": 2.51039501039501,
      "grad_norm": 0.29023870825767517,
      "learning_rate": 5.534765534765535e-06,
      "loss": 0.0064,
      "step": 2415
    },
    {
      "epoch": 2.5114345114345116,
      "grad_norm": 0.6130965352058411,
      "learning_rate": 5.532455532455533e-06,
      "loss": 0.016,
      "step": 2416
    },
    {
      "epoch": 2.5124740124740126,
      "grad_norm": 12.549509048461914,
      "learning_rate": 5.530145530145531e-06,
      "loss": 0.7434,
      "step": 2417
    },
    {
      "epoch": 2.5135135135135136,
      "grad_norm": 11.806109428405762,
      "learning_rate": 5.527835527835528e-06,
      "loss": 0.8109,
      "step": 2418
    },
    {
      "epoch": 2.5145530145530146,
      "grad_norm": 0.16034944355487823,
      "learning_rate": 5.5255255255255255e-06,
      "loss": 0.003,
      "step": 2419
    },
    {
      "epoch": 2.5155925155925156,
      "grad_norm": 1.2844282388687134,
      "learning_rate": 5.523215523215524e-06,
      "loss": 0.0154,
      "step": 2420
    },
    {
      "epoch": 2.5166320166320166,
      "grad_norm": 6.30752420425415,
      "learning_rate": 5.520905520905522e-06,
      "loss": 0.1711,
      "step": 2421
    },
    {
      "epoch": 2.5176715176715176,
      "grad_norm": 9.42320728302002,
      "learning_rate": 5.518595518595518e-06,
      "loss": 0.5665,
      "step": 2422
    },
    {
      "epoch": 2.5187110187110187,
      "grad_norm": 8.125027656555176,
      "learning_rate": 5.516285516285517e-06,
      "loss": 0.5858,
      "step": 2423
    },
    {
      "epoch": 2.5197505197505197,
      "grad_norm": 11.059926986694336,
      "learning_rate": 5.513975513975515e-06,
      "loss": 0.7028,
      "step": 2424
    },
    {
      "epoch": 2.5207900207900207,
      "grad_norm": 1.8393372297286987,
      "learning_rate": 5.511665511665513e-06,
      "loss": 0.0419,
      "step": 2425
    },
    {
      "epoch": 2.5218295218295217,
      "grad_norm": 1.991614818572998,
      "learning_rate": 5.5093555093555095e-06,
      "loss": 0.0236,
      "step": 2426
    },
    {
      "epoch": 2.5228690228690227,
      "grad_norm": 0.1596309095621109,
      "learning_rate": 5.507045507045508e-06,
      "loss": 0.0015,
      "step": 2427
    },
    {
      "epoch": 2.523908523908524,
      "grad_norm": 3.80548095703125,
      "learning_rate": 5.504735504735506e-06,
      "loss": 0.1103,
      "step": 2428
    },
    {
      "epoch": 2.524948024948025,
      "grad_norm": 0.8834503889083862,
      "learning_rate": 5.502425502425502e-06,
      "loss": 0.2529,
      "step": 2429
    },
    {
      "epoch": 2.525987525987526,
      "grad_norm": 1.7620978355407715,
      "learning_rate": 5.5001155001155005e-06,
      "loss": 0.0571,
      "step": 2430
    },
    {
      "epoch": 2.527027027027027,
      "grad_norm": 8.883015632629395,
      "learning_rate": 5.497805497805499e-06,
      "loss": 0.2079,
      "step": 2431
    },
    {
      "epoch": 2.528066528066528,
      "grad_norm": 0.02310093306005001,
      "learning_rate": 5.495495495495496e-06,
      "loss": 0.0006,
      "step": 2432
    },
    {
      "epoch": 2.529106029106029,
      "grad_norm": 0.1949092447757721,
      "learning_rate": 5.493185493185493e-06,
      "loss": 0.0058,
      "step": 2433
    },
    {
      "epoch": 2.5301455301455302,
      "grad_norm": 0.004292343743145466,
      "learning_rate": 5.490875490875492e-06,
      "loss": 0.0,
      "step": 2434
    },
    {
      "epoch": 2.5311850311850312,
      "grad_norm": 6.701157093048096,
      "learning_rate": 5.488565488565489e-06,
      "loss": 0.4612,
      "step": 2435
    },
    {
      "epoch": 2.5322245322245323,
      "grad_norm": 2.1965763568878174,
      "learning_rate": 5.486255486255486e-06,
      "loss": 0.0605,
      "step": 2436
    },
    {
      "epoch": 2.5332640332640333,
      "grad_norm": 5.151659965515137,
      "learning_rate": 5.4839454839454845e-06,
      "loss": 0.135,
      "step": 2437
    },
    {
      "epoch": 2.5343035343035343,
      "grad_norm": 0.12913161516189575,
      "learning_rate": 5.481635481635482e-06,
      "loss": 0.0034,
      "step": 2438
    },
    {
      "epoch": 2.5353430353430353,
      "grad_norm": 3.0704658031463623,
      "learning_rate": 5.47932547932548e-06,
      "loss": 0.0782,
      "step": 2439
    },
    {
      "epoch": 2.5363825363825363,
      "grad_norm": 1.6351587772369385,
      "learning_rate": 5.477015477015477e-06,
      "loss": 0.1708,
      "step": 2440
    },
    {
      "epoch": 2.5374220374220373,
      "grad_norm": 0.03910132497549057,
      "learning_rate": 5.474705474705475e-06,
      "loss": 0.0012,
      "step": 2441
    },
    {
      "epoch": 2.5384615384615383,
      "grad_norm": 3.443721294403076,
      "learning_rate": 5.472395472395473e-06,
      "loss": 0.0967,
      "step": 2442
    },
    {
      "epoch": 2.5395010395010393,
      "grad_norm": 3.9343323707580566,
      "learning_rate": 5.470085470085471e-06,
      "loss": 0.1029,
      "step": 2443
    },
    {
      "epoch": 2.5405405405405403,
      "grad_norm": 8.639856338500977,
      "learning_rate": 5.467775467775468e-06,
      "loss": 0.6861,
      "step": 2444
    },
    {
      "epoch": 2.5415800415800414,
      "grad_norm": 3.257338047027588,
      "learning_rate": 5.465465465465466e-06,
      "loss": 0.1017,
      "step": 2445
    },
    {
      "epoch": 2.5426195426195424,
      "grad_norm": 0.10939653217792511,
      "learning_rate": 5.463155463155464e-06,
      "loss": 0.0017,
      "step": 2446
    },
    {
      "epoch": 2.5436590436590434,
      "grad_norm": 0.6103467345237732,
      "learning_rate": 5.4608454608454605e-06,
      "loss": 0.0167,
      "step": 2447
    },
    {
      "epoch": 2.544698544698545,
      "grad_norm": 0.9619538187980652,
      "learning_rate": 5.458535458535459e-06,
      "loss": 0.015,
      "step": 2448
    },
    {
      "epoch": 2.545738045738046,
      "grad_norm": 12.676862716674805,
      "learning_rate": 5.456225456225457e-06,
      "loss": 1.0984,
      "step": 2449
    },
    {
      "epoch": 2.546777546777547,
      "grad_norm": 13.927125930786133,
      "learning_rate": 5.453915453915455e-06,
      "loss": 1.5171,
      "step": 2450
    },
    {
      "epoch": 2.547817047817048,
      "grad_norm": 8.507790565490723,
      "learning_rate": 5.4516054516054515e-06,
      "loss": 0.2561,
      "step": 2451
    },
    {
      "epoch": 2.548856548856549,
      "grad_norm": 10.253948211669922,
      "learning_rate": 5.44929544929545e-06,
      "loss": 0.6676,
      "step": 2452
    },
    {
      "epoch": 2.54989604989605,
      "grad_norm": 3.853451728820801,
      "learning_rate": 5.446985446985448e-06,
      "loss": 0.1021,
      "step": 2453
    },
    {
      "epoch": 2.550935550935551,
      "grad_norm": 7.904906749725342,
      "learning_rate": 5.444675444675444e-06,
      "loss": 1.0231,
      "step": 2454
    },
    {
      "epoch": 2.551975051975052,
      "grad_norm": 10.921005249023438,
      "learning_rate": 5.442365442365443e-06,
      "loss": 0.4457,
      "step": 2455
    },
    {
      "epoch": 2.553014553014553,
      "grad_norm": 7.966489315032959,
      "learning_rate": 5.440055440055441e-06,
      "loss": 0.8084,
      "step": 2456
    },
    {
      "epoch": 2.554054054054054,
      "grad_norm": 0.3931668698787689,
      "learning_rate": 5.437745437745439e-06,
      "loss": 0.0101,
      "step": 2457
    },
    {
      "epoch": 2.555093555093555,
      "grad_norm": 0.6197918653488159,
      "learning_rate": 5.4354354354354355e-06,
      "loss": 0.0187,
      "step": 2458
    },
    {
      "epoch": 2.556133056133056,
      "grad_norm": 5.144542217254639,
      "learning_rate": 5.433125433125434e-06,
      "loss": 0.1461,
      "step": 2459
    },
    {
      "epoch": 2.5571725571725574,
      "grad_norm": 0.13302533328533173,
      "learning_rate": 5.430815430815432e-06,
      "loss": 0.0023,
      "step": 2460
    },
    {
      "epoch": 2.5582120582120584,
      "grad_norm": 0.4320324659347534,
      "learning_rate": 5.428505428505429e-06,
      "loss": 0.0132,
      "step": 2461
    },
    {
      "epoch": 2.5592515592515594,
      "grad_norm": 11.432194709777832,
      "learning_rate": 5.4261954261954265e-06,
      "loss": 1.0018,
      "step": 2462
    },
    {
      "epoch": 2.5602910602910605,
      "grad_norm": 6.14457893371582,
      "learning_rate": 5.423885423885425e-06,
      "loss": 0.2512,
      "step": 2463
    },
    {
      "epoch": 2.5613305613305615,
      "grad_norm": 10.37625503540039,
      "learning_rate": 5.421575421575422e-06,
      "loss": 0.8207,
      "step": 2464
    },
    {
      "epoch": 2.5623700623700625,
      "grad_norm": 7.509644508361816,
      "learning_rate": 5.4192654192654194e-06,
      "loss": 1.0264,
      "step": 2465
    },
    {
      "epoch": 2.5634095634095635,
      "grad_norm": 14.384940147399902,
      "learning_rate": 5.416955416955418e-06,
      "loss": 0.8239,
      "step": 2466
    },
    {
      "epoch": 2.5644490644490645,
      "grad_norm": 6.205191612243652,
      "learning_rate": 5.414645414645415e-06,
      "loss": 0.1594,
      "step": 2467
    },
    {
      "epoch": 2.5654885654885655,
      "grad_norm": 9.01181411743164,
      "learning_rate": 5.412335412335413e-06,
      "loss": 0.297,
      "step": 2468
    },
    {
      "epoch": 2.5665280665280665,
      "grad_norm": 1.9883480072021484,
      "learning_rate": 5.4100254100254105e-06,
      "loss": 0.0604,
      "step": 2469
    },
    {
      "epoch": 2.5675675675675675,
      "grad_norm": 0.9856329560279846,
      "learning_rate": 5.407715407715408e-06,
      "loss": 0.0192,
      "step": 2470
    },
    {
      "epoch": 2.5686070686070686,
      "grad_norm": 0.8478329181671143,
      "learning_rate": 5.405405405405406e-06,
      "loss": 0.0229,
      "step": 2471
    },
    {
      "epoch": 2.5696465696465696,
      "grad_norm": 0.5473334789276123,
      "learning_rate": 5.403095403095403e-06,
      "loss": 0.0162,
      "step": 2472
    },
    {
      "epoch": 2.5706860706860706,
      "grad_norm": 11.216007232666016,
      "learning_rate": 5.400785400785401e-06,
      "loss": 0.6939,
      "step": 2473
    },
    {
      "epoch": 2.5717255717255716,
      "grad_norm": 0.467229962348938,
      "learning_rate": 5.398475398475399e-06,
      "loss": 0.0103,
      "step": 2474
    },
    {
      "epoch": 2.5727650727650726,
      "grad_norm": 1.6031438112258911,
      "learning_rate": 5.396165396165397e-06,
      "loss": 0.0448,
      "step": 2475
    },
    {
      "epoch": 2.5738045738045736,
      "grad_norm": 8.846580505371094,
      "learning_rate": 5.393855393855394e-06,
      "loss": 0.7402,
      "step": 2476
    },
    {
      "epoch": 2.5748440748440746,
      "grad_norm": 4.420310020446777,
      "learning_rate": 5.391545391545392e-06,
      "loss": 0.1099,
      "step": 2477
    },
    {
      "epoch": 2.5758835758835756,
      "grad_norm": 9.618853569030762,
      "learning_rate": 5.38923538923539e-06,
      "loss": 0.643,
      "step": 2478
    },
    {
      "epoch": 2.5769230769230766,
      "grad_norm": 8.873306274414062,
      "learning_rate": 5.386925386925388e-06,
      "loss": 0.7841,
      "step": 2479
    },
    {
      "epoch": 2.577962577962578,
      "grad_norm": 18.581398010253906,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1.7637,
      "step": 2480
    },
    {
      "epoch": 2.579002079002079,
      "grad_norm": 10.233980178833008,
      "learning_rate": 5.382305382305383e-06,
      "loss": 0.764,
      "step": 2481
    },
    {
      "epoch": 2.58004158004158,
      "grad_norm": 1.2297041416168213,
      "learning_rate": 5.379995379995381e-06,
      "loss": 0.0358,
      "step": 2482
    },
    {
      "epoch": 2.581081081081081,
      "grad_norm": 6.728565216064453,
      "learning_rate": 5.3776853776853775e-06,
      "loss": 0.5458,
      "step": 2483
    },
    {
      "epoch": 2.582120582120582,
      "grad_norm": 3.641315221786499,
      "learning_rate": 5.375375375375376e-06,
      "loss": 0.04,
      "step": 2484
    },
    {
      "epoch": 2.583160083160083,
      "grad_norm": 0.34038031101226807,
      "learning_rate": 5.373065373065374e-06,
      "loss": 0.0113,
      "step": 2485
    },
    {
      "epoch": 2.584199584199584,
      "grad_norm": 0.9422138333320618,
      "learning_rate": 5.370755370755371e-06,
      "loss": 0.0299,
      "step": 2486
    },
    {
      "epoch": 2.585239085239085,
      "grad_norm": 0.12205638736486435,
      "learning_rate": 5.368445368445369e-06,
      "loss": 0.0042,
      "step": 2487
    },
    {
      "epoch": 2.586278586278586,
      "grad_norm": 7.378511905670166,
      "learning_rate": 5.366135366135367e-06,
      "loss": 0.2885,
      "step": 2488
    },
    {
      "epoch": 2.587318087318087,
      "grad_norm": 0.26321983337402344,
      "learning_rate": 5.363825363825364e-06,
      "loss": 0.007,
      "step": 2489
    },
    {
      "epoch": 2.5883575883575882,
      "grad_norm": 6.9181342124938965,
      "learning_rate": 5.3615153615153615e-06,
      "loss": 0.2783,
      "step": 2490
    },
    {
      "epoch": 2.5893970893970892,
      "grad_norm": 0.12993164360523224,
      "learning_rate": 5.35920535920536e-06,
      "loss": 0.0037,
      "step": 2491
    },
    {
      "epoch": 2.5904365904365907,
      "grad_norm": 0.17028915882110596,
      "learning_rate": 5.356895356895357e-06,
      "loss": 0.0057,
      "step": 2492
    },
    {
      "epoch": 2.5914760914760917,
      "grad_norm": 5.357726573944092,
      "learning_rate": 5.354585354585355e-06,
      "loss": 0.553,
      "step": 2493
    },
    {
      "epoch": 2.5925155925155927,
      "grad_norm": 0.08897285163402557,
      "learning_rate": 5.3522753522753526e-06,
      "loss": 0.0022,
      "step": 2494
    },
    {
      "epoch": 2.5935550935550937,
      "grad_norm": 9.253763198852539,
      "learning_rate": 5.349965349965351e-06,
      "loss": 0.6422,
      "step": 2495
    },
    {
      "epoch": 2.5945945945945947,
      "grad_norm": 0.5697972774505615,
      "learning_rate": 5.347655347655348e-06,
      "loss": 0.0165,
      "step": 2496
    },
    {
      "epoch": 2.5956340956340958,
      "grad_norm": 4.335848331451416,
      "learning_rate": 5.345345345345346e-06,
      "loss": 0.2975,
      "step": 2497
    },
    {
      "epoch": 2.5966735966735968,
      "grad_norm": 4.842863082885742,
      "learning_rate": 5.343035343035344e-06,
      "loss": 0.1894,
      "step": 2498
    },
    {
      "epoch": 2.5977130977130978,
      "grad_norm": 0.643163800239563,
      "learning_rate": 5.340725340725341e-06,
      "loss": 0.0129,
      "step": 2499
    },
    {
      "epoch": 2.598752598752599,
      "grad_norm": 6.766790390014648,
      "learning_rate": 5.338415338415339e-06,
      "loss": 0.4936,
      "step": 2500
    },
    {
      "epoch": 2.5997920997921,
      "grad_norm": 0.04399247094988823,
      "learning_rate": 5.3361053361053365e-06,
      "loss": 0.0013,
      "step": 2501
    },
    {
      "epoch": 2.600831600831601,
      "grad_norm": 9.183923721313477,
      "learning_rate": 5.333795333795334e-06,
      "loss": 0.4783,
      "step": 2502
    },
    {
      "epoch": 2.601871101871102,
      "grad_norm": 4.868300437927246,
      "learning_rate": 5.331485331485332e-06,
      "loss": 0.3321,
      "step": 2503
    },
    {
      "epoch": 2.602910602910603,
      "grad_norm": 7.282240867614746,
      "learning_rate": 5.32917532917533e-06,
      "loss": 0.5301,
      "step": 2504
    },
    {
      "epoch": 2.603950103950104,
      "grad_norm": 3.4226865768432617,
      "learning_rate": 5.326865326865327e-06,
      "loss": 0.1984,
      "step": 2505
    },
    {
      "epoch": 2.604989604989605,
      "grad_norm": 4.259052276611328,
      "learning_rate": 5.324555324555325e-06,
      "loss": 0.2619,
      "step": 2506
    },
    {
      "epoch": 2.606029106029106,
      "grad_norm": 6.480755805969238,
      "learning_rate": 5.322245322245323e-06,
      "loss": 0.3234,
      "step": 2507
    },
    {
      "epoch": 2.607068607068607,
      "grad_norm": 0.06338009238243103,
      "learning_rate": 5.31993531993532e-06,
      "loss": 0.0018,
      "step": 2508
    },
    {
      "epoch": 2.608108108108108,
      "grad_norm": 3.376009464263916,
      "learning_rate": 5.317625317625318e-06,
      "loss": 0.1565,
      "step": 2509
    },
    {
      "epoch": 2.609147609147609,
      "grad_norm": 2.3504629135131836,
      "learning_rate": 5.315315315315316e-06,
      "loss": 0.0513,
      "step": 2510
    },
    {
      "epoch": 2.61018711018711,
      "grad_norm": 3.9377503395080566,
      "learning_rate": 5.313005313005314e-06,
      "loss": 0.3037,
      "step": 2511
    },
    {
      "epoch": 2.6112266112266114,
      "grad_norm": 2.743152141571045,
      "learning_rate": 5.310695310695311e-06,
      "loss": 0.1154,
      "step": 2512
    },
    {
      "epoch": 2.6122661122661124,
      "grad_norm": 1.6104410886764526,
      "learning_rate": 5.308385308385309e-06,
      "loss": 0.0512,
      "step": 2513
    },
    {
      "epoch": 2.6133056133056134,
      "grad_norm": 0.4035050868988037,
      "learning_rate": 5.306075306075307e-06,
      "loss": 0.0109,
      "step": 2514
    },
    {
      "epoch": 2.6143451143451144,
      "grad_norm": 1.360506534576416,
      "learning_rate": 5.303765303765304e-06,
      "loss": 0.0475,
      "step": 2515
    },
    {
      "epoch": 2.6153846153846154,
      "grad_norm": 9.4105863571167,
      "learning_rate": 5.301455301455302e-06,
      "loss": 1.3063,
      "step": 2516
    },
    {
      "epoch": 2.6164241164241164,
      "grad_norm": 5.020283222198486,
      "learning_rate": 5.2991452991453e-06,
      "loss": 0.2818,
      "step": 2517
    },
    {
      "epoch": 2.6174636174636174,
      "grad_norm": 0.8245450258255005,
      "learning_rate": 5.296835296835297e-06,
      "loss": 0.0157,
      "step": 2518
    },
    {
      "epoch": 2.6185031185031185,
      "grad_norm": 5.981417655944824,
      "learning_rate": 5.294525294525295e-06,
      "loss": 0.1196,
      "step": 2519
    },
    {
      "epoch": 2.6195426195426195,
      "grad_norm": 1.7131268978118896,
      "learning_rate": 5.292215292215293e-06,
      "loss": 0.0433,
      "step": 2520
    },
    {
      "epoch": 2.6205821205821205,
      "grad_norm": 8.64378833770752,
      "learning_rate": 5.28990528990529e-06,
      "loss": 0.5102,
      "step": 2521
    },
    {
      "epoch": 2.6216216216216215,
      "grad_norm": 11.704745292663574,
      "learning_rate": 5.287595287595288e-06,
      "loss": 0.6297,
      "step": 2522
    },
    {
      "epoch": 2.6226611226611225,
      "grad_norm": 8.039158821105957,
      "learning_rate": 5.285285285285286e-06,
      "loss": 0.3796,
      "step": 2523
    },
    {
      "epoch": 2.623700623700624,
      "grad_norm": 13.407707214355469,
      "learning_rate": 5.282975282975283e-06,
      "loss": 0.5751,
      "step": 2524
    },
    {
      "epoch": 2.624740124740125,
      "grad_norm": 4.56635046005249,
      "learning_rate": 5.280665280665281e-06,
      "loss": 0.2088,
      "step": 2525
    },
    {
      "epoch": 2.625779625779626,
      "grad_norm": 5.892931938171387,
      "learning_rate": 5.2783552783552786e-06,
      "loss": 0.2088,
      "step": 2526
    },
    {
      "epoch": 2.626819126819127,
      "grad_norm": 5.208812713623047,
      "learning_rate": 5.276045276045276e-06,
      "loss": 0.3068,
      "step": 2527
    },
    {
      "epoch": 2.627858627858628,
      "grad_norm": 1.0516034364700317,
      "learning_rate": 5.273735273735274e-06,
      "loss": 0.0139,
      "step": 2528
    },
    {
      "epoch": 2.628898128898129,
      "grad_norm": 6.1416239738464355,
      "learning_rate": 5.271425271425272e-06,
      "loss": 0.3791,
      "step": 2529
    },
    {
      "epoch": 2.62993762993763,
      "grad_norm": 4.7939863204956055,
      "learning_rate": 5.269115269115269e-06,
      "loss": 0.3712,
      "step": 2530
    },
    {
      "epoch": 2.630977130977131,
      "grad_norm": 3.7652533054351807,
      "learning_rate": 5.266805266805267e-06,
      "loss": 0.1008,
      "step": 2531
    },
    {
      "epoch": 2.632016632016632,
      "grad_norm": 7.166833877563477,
      "learning_rate": 5.264495264495265e-06,
      "loss": 0.4992,
      "step": 2532
    },
    {
      "epoch": 2.633056133056133,
      "grad_norm": 8.62056827545166,
      "learning_rate": 5.262185262185263e-06,
      "loss": 0.3708,
      "step": 2533
    },
    {
      "epoch": 2.634095634095634,
      "grad_norm": 2.3554868698120117,
      "learning_rate": 5.25987525987526e-06,
      "loss": 0.0478,
      "step": 2534
    },
    {
      "epoch": 2.635135135135135,
      "grad_norm": 8.775339126586914,
      "learning_rate": 5.257565257565258e-06,
      "loss": 0.3151,
      "step": 2535
    },
    {
      "epoch": 2.636174636174636,
      "grad_norm": 2.1053552627563477,
      "learning_rate": 5.255255255255256e-06,
      "loss": 0.2953,
      "step": 2536
    },
    {
      "epoch": 2.637214137214137,
      "grad_norm": 4.951183795928955,
      "learning_rate": 5.252945252945253e-06,
      "loss": 0.1592,
      "step": 2537
    },
    {
      "epoch": 2.638253638253638,
      "grad_norm": 0.02089507319033146,
      "learning_rate": 5.250635250635251e-06,
      "loss": 0.0005,
      "step": 2538
    },
    {
      "epoch": 2.639293139293139,
      "grad_norm": 0.20281526446342468,
      "learning_rate": 5.248325248325249e-06,
      "loss": 0.0041,
      "step": 2539
    },
    {
      "epoch": 2.64033264033264,
      "grad_norm": 7.6722002029418945,
      "learning_rate": 5.246015246015247e-06,
      "loss": 0.4628,
      "step": 2540
    },
    {
      "epoch": 2.641372141372141,
      "grad_norm": 9.445036888122559,
      "learning_rate": 5.243705243705244e-06,
      "loss": 0.8533,
      "step": 2541
    },
    {
      "epoch": 2.642411642411642,
      "grad_norm": 0.6576030254364014,
      "learning_rate": 5.241395241395242e-06,
      "loss": 0.0151,
      "step": 2542
    },
    {
      "epoch": 2.643451143451143,
      "grad_norm": 5.462561130523682,
      "learning_rate": 5.23908523908524e-06,
      "loss": 0.1482,
      "step": 2543
    },
    {
      "epoch": 2.6444906444906446,
      "grad_norm": 0.2012774497270584,
      "learning_rate": 5.236775236775237e-06,
      "loss": 0.0048,
      "step": 2544
    },
    {
      "epoch": 2.6455301455301456,
      "grad_norm": 6.6424174308776855,
      "learning_rate": 5.234465234465235e-06,
      "loss": 0.2679,
      "step": 2545
    },
    {
      "epoch": 2.6465696465696467,
      "grad_norm": 7.657466411590576,
      "learning_rate": 5.232155232155233e-06,
      "loss": 0.2041,
      "step": 2546
    },
    {
      "epoch": 2.6476091476091477,
      "grad_norm": 0.44189453125,
      "learning_rate": 5.22984522984523e-06,
      "loss": 0.013,
      "step": 2547
    },
    {
      "epoch": 2.6486486486486487,
      "grad_norm": 1.6354267597198486,
      "learning_rate": 5.227535227535228e-06,
      "loss": 0.0265,
      "step": 2548
    },
    {
      "epoch": 2.6496881496881497,
      "grad_norm": 4.967313766479492,
      "learning_rate": 5.225225225225226e-06,
      "loss": 0.1779,
      "step": 2549
    },
    {
      "epoch": 2.6507276507276507,
      "grad_norm": 0.6553028225898743,
      "learning_rate": 5.222915222915223e-06,
      "loss": 0.0178,
      "step": 2550
    },
    {
      "epoch": 2.6517671517671517,
      "grad_norm": 5.932888031005859,
      "learning_rate": 5.2206052206052215e-06,
      "loss": 0.2465,
      "step": 2551
    },
    {
      "epoch": 2.6528066528066527,
      "grad_norm": 1.0269923210144043,
      "learning_rate": 5.218295218295219e-06,
      "loss": 0.0224,
      "step": 2552
    },
    {
      "epoch": 2.6538461538461537,
      "grad_norm": 4.396491050720215,
      "learning_rate": 5.215985215985216e-06,
      "loss": 0.1572,
      "step": 2553
    },
    {
      "epoch": 2.6548856548856548,
      "grad_norm": 0.5629084706306458,
      "learning_rate": 5.213675213675214e-06,
      "loss": 0.0216,
      "step": 2554
    },
    {
      "epoch": 2.6559251559251558,
      "grad_norm": 0.1629319041967392,
      "learning_rate": 5.211365211365212e-06,
      "loss": 0.0029,
      "step": 2555
    },
    {
      "epoch": 2.6569646569646572,
      "grad_norm": 8.487727165222168,
      "learning_rate": 5.209055209055209e-06,
      "loss": 1.2105,
      "step": 2556
    },
    {
      "epoch": 2.6580041580041582,
      "grad_norm": 6.604908466339111,
      "learning_rate": 5.206745206745207e-06,
      "loss": 0.3326,
      "step": 2557
    },
    {
      "epoch": 2.6590436590436592,
      "grad_norm": 9.358423233032227,
      "learning_rate": 5.2044352044352054e-06,
      "loss": 0.6385,
      "step": 2558
    },
    {
      "epoch": 2.6600831600831603,
      "grad_norm": 0.8450285196304321,
      "learning_rate": 5.202125202125202e-06,
      "loss": 0.013,
      "step": 2559
    },
    {
      "epoch": 2.6611226611226613,
      "grad_norm": 8.273629188537598,
      "learning_rate": 5.1998151998152e-06,
      "loss": 0.2557,
      "step": 2560
    },
    {
      "epoch": 2.6621621621621623,
      "grad_norm": 0.030706413090229034,
      "learning_rate": 5.197505197505198e-06,
      "loss": 0.0007,
      "step": 2561
    },
    {
      "epoch": 2.6632016632016633,
      "grad_norm": 3.37729811668396,
      "learning_rate": 5.195195195195195e-06,
      "loss": 0.1005,
      "step": 2562
    },
    {
      "epoch": 2.6642411642411643,
      "grad_norm": 0.17317216098308563,
      "learning_rate": 5.192885192885193e-06,
      "loss": 0.0039,
      "step": 2563
    },
    {
      "epoch": 2.6652806652806653,
      "grad_norm": 3.6512608528137207,
      "learning_rate": 5.190575190575191e-06,
      "loss": 0.1058,
      "step": 2564
    },
    {
      "epoch": 2.6663201663201663,
      "grad_norm": 2.5752034187316895,
      "learning_rate": 5.188265188265189e-06,
      "loss": 0.1316,
      "step": 2565
    },
    {
      "epoch": 2.6673596673596673,
      "grad_norm": 1.0415226221084595,
      "learning_rate": 5.185955185955186e-06,
      "loss": 0.0336,
      "step": 2566
    },
    {
      "epoch": 2.6683991683991684,
      "grad_norm": 3.302722454071045,
      "learning_rate": 5.183645183645184e-06,
      "loss": 0.0712,
      "step": 2567
    },
    {
      "epoch": 2.6694386694386694,
      "grad_norm": 1.884563684463501,
      "learning_rate": 5.181335181335182e-06,
      "loss": 0.1313,
      "step": 2568
    },
    {
      "epoch": 2.6704781704781704,
      "grad_norm": 1.5982835292816162,
      "learning_rate": 5.17902517902518e-06,
      "loss": 0.0511,
      "step": 2569
    },
    {
      "epoch": 2.6715176715176714,
      "grad_norm": 5.131686687469482,
      "learning_rate": 5.176715176715177e-06,
      "loss": 0.2519,
      "step": 2570
    },
    {
      "epoch": 2.6725571725571724,
      "grad_norm": 7.3811564445495605,
      "learning_rate": 5.174405174405175e-06,
      "loss": 0.4257,
      "step": 2571
    },
    {
      "epoch": 2.6735966735966734,
      "grad_norm": 0.13512037694454193,
      "learning_rate": 5.1720951720951725e-06,
      "loss": 0.0034,
      "step": 2572
    },
    {
      "epoch": 2.6746361746361744,
      "grad_norm": 3.146843194961548,
      "learning_rate": 5.16978516978517e-06,
      "loss": 0.123,
      "step": 2573
    },
    {
      "epoch": 2.6756756756756754,
      "grad_norm": 0.004672181326895952,
      "learning_rate": 5.167475167475168e-06,
      "loss": 0.0001,
      "step": 2574
    },
    {
      "epoch": 2.6767151767151764,
      "grad_norm": 1.4410141706466675,
      "learning_rate": 5.165165165165165e-06,
      "loss": 0.0345,
      "step": 2575
    },
    {
      "epoch": 2.677754677754678,
      "grad_norm": 0.6777349710464478,
      "learning_rate": 5.1628551628551635e-06,
      "loss": 0.0282,
      "step": 2576
    },
    {
      "epoch": 2.678794178794179,
      "grad_norm": 0.03477311134338379,
      "learning_rate": 5.160545160545161e-06,
      "loss": 0.0012,
      "step": 2577
    },
    {
      "epoch": 2.67983367983368,
      "grad_norm": 4.510110378265381,
      "learning_rate": 5.158235158235158e-06,
      "loss": 0.0668,
      "step": 2578
    },
    {
      "epoch": 2.680873180873181,
      "grad_norm": 13.6720609664917,
      "learning_rate": 5.155925155925156e-06,
      "loss": 2.3046,
      "step": 2579
    },
    {
      "epoch": 2.681912681912682,
      "grad_norm": 0.8042404055595398,
      "learning_rate": 5.153615153615154e-06,
      "loss": 0.0317,
      "step": 2580
    },
    {
      "epoch": 2.682952182952183,
      "grad_norm": 2.2485601902008057,
      "learning_rate": 5.151305151305151e-06,
      "loss": 0.0665,
      "step": 2581
    },
    {
      "epoch": 2.683991683991684,
      "grad_norm": 0.08799292892217636,
      "learning_rate": 5.148995148995149e-06,
      "loss": 0.0025,
      "step": 2582
    },
    {
      "epoch": 2.685031185031185,
      "grad_norm": 2.3404698371887207,
      "learning_rate": 5.1466851466851475e-06,
      "loss": 0.0433,
      "step": 2583
    },
    {
      "epoch": 2.686070686070686,
      "grad_norm": 0.5272285342216492,
      "learning_rate": 5.144375144375144e-06,
      "loss": 0.0118,
      "step": 2584
    },
    {
      "epoch": 2.687110187110187,
      "grad_norm": 0.020404517650604248,
      "learning_rate": 5.142065142065142e-06,
      "loss": 0.0006,
      "step": 2585
    },
    {
      "epoch": 2.688149688149688,
      "grad_norm": 0.5791797041893005,
      "learning_rate": 5.13975513975514e-06,
      "loss": 0.0176,
      "step": 2586
    },
    {
      "epoch": 2.689189189189189,
      "grad_norm": 7.628298282623291,
      "learning_rate": 5.1374451374451386e-06,
      "loss": 0.3845,
      "step": 2587
    },
    {
      "epoch": 2.6902286902286905,
      "grad_norm": 3.025831460952759,
      "learning_rate": 5.135135135135135e-06,
      "loss": 0.2582,
      "step": 2588
    },
    {
      "epoch": 2.6912681912681915,
      "grad_norm": 6.124800682067871,
      "learning_rate": 5.132825132825133e-06,
      "loss": 0.1955,
      "step": 2589
    },
    {
      "epoch": 2.6923076923076925,
      "grad_norm": 2.600257158279419,
      "learning_rate": 5.1305151305151314e-06,
      "loss": 0.0648,
      "step": 2590
    },
    {
      "epoch": 2.6933471933471935,
      "grad_norm": 4.131895542144775,
      "learning_rate": 5.128205128205128e-06,
      "loss": 0.097,
      "step": 2591
    },
    {
      "epoch": 2.6943866943866945,
      "grad_norm": 0.0008171483059413731,
      "learning_rate": 5.125895125895126e-06,
      "loss": 0.0,
      "step": 2592
    },
    {
      "epoch": 2.6954261954261955,
      "grad_norm": 14.881963729858398,
      "learning_rate": 5.123585123585124e-06,
      "loss": 1.7013,
      "step": 2593
    },
    {
      "epoch": 2.6964656964656966,
      "grad_norm": 2.45265531539917,
      "learning_rate": 5.1212751212751225e-06,
      "loss": 0.0527,
      "step": 2594
    },
    {
      "epoch": 2.6975051975051976,
      "grad_norm": 11.552932739257812,
      "learning_rate": 5.118965118965119e-06,
      "loss": 0.6112,
      "step": 2595
    },
    {
      "epoch": 2.6985446985446986,
      "grad_norm": 0.15990565717220306,
      "learning_rate": 5.116655116655117e-06,
      "loss": 0.0037,
      "step": 2596
    },
    {
      "epoch": 2.6995841995841996,
      "grad_norm": 0.015272875316441059,
      "learning_rate": 5.114345114345115e-06,
      "loss": 0.0003,
      "step": 2597
    },
    {
      "epoch": 2.7006237006237006,
      "grad_norm": 10.404386520385742,
      "learning_rate": 5.112035112035112e-06,
      "loss": 1.4427,
      "step": 2598
    },
    {
      "epoch": 2.7016632016632016,
      "grad_norm": 3.653311252593994,
      "learning_rate": 5.10972510972511e-06,
      "loss": 0.1365,
      "step": 2599
    },
    {
      "epoch": 2.7027027027027026,
      "grad_norm": 6.013213634490967,
      "learning_rate": 5.107415107415108e-06,
      "loss": 0.1969,
      "step": 2600
    },
    {
      "epoch": 2.7037422037422036,
      "grad_norm": 0.7424914240837097,
      "learning_rate": 5.105105105105106e-06,
      "loss": 0.0127,
      "step": 2601
    },
    {
      "epoch": 2.7047817047817047,
      "grad_norm": 0.16230179369449615,
      "learning_rate": 5.102795102795103e-06,
      "loss": 0.0046,
      "step": 2602
    },
    {
      "epoch": 2.7058212058212057,
      "grad_norm": 8.930994033813477,
      "learning_rate": 5.100485100485101e-06,
      "loss": 0.5425,
      "step": 2603
    },
    {
      "epoch": 2.7068607068607067,
      "grad_norm": 1.592777132987976,
      "learning_rate": 5.0981750981750985e-06,
      "loss": 0.029,
      "step": 2604
    },
    {
      "epoch": 2.7079002079002077,
      "grad_norm": 2.0030179023742676,
      "learning_rate": 5.095865095865097e-06,
      "loss": 0.042,
      "step": 2605
    },
    {
      "epoch": 2.7089397089397087,
      "grad_norm": 4.972201824188232,
      "learning_rate": 5.093555093555094e-06,
      "loss": 0.2896,
      "step": 2606
    },
    {
      "epoch": 2.7099792099792097,
      "grad_norm": 0.016869381070137024,
      "learning_rate": 5.091245091245091e-06,
      "loss": 0.0004,
      "step": 2607
    },
    {
      "epoch": 2.711018711018711,
      "grad_norm": 2.6783928871154785,
      "learning_rate": 5.0889350889350895e-06,
      "loss": 0.1094,
      "step": 2608
    },
    {
      "epoch": 2.712058212058212,
      "grad_norm": 11.518509864807129,
      "learning_rate": 5.086625086625087e-06,
      "loss": 0.5764,
      "step": 2609
    },
    {
      "epoch": 2.713097713097713,
      "grad_norm": 2.1756484508514404,
      "learning_rate": 5.084315084315084e-06,
      "loss": 0.0298,
      "step": 2610
    },
    {
      "epoch": 2.714137214137214,
      "grad_norm": 2.0496127605438232,
      "learning_rate": 5.0820050820050824e-06,
      "loss": 0.034,
      "step": 2611
    },
    {
      "epoch": 2.715176715176715,
      "grad_norm": 5.0031561851501465,
      "learning_rate": 5.079695079695081e-06,
      "loss": 0.1148,
      "step": 2612
    },
    {
      "epoch": 2.7162162162162162,
      "grad_norm": 3.1190743446350098,
      "learning_rate": 5.077385077385077e-06,
      "loss": 0.0171,
      "step": 2613
    },
    {
      "epoch": 2.7172557172557172,
      "grad_norm": 2.5663068294525146,
      "learning_rate": 5.075075075075075e-06,
      "loss": 0.0828,
      "step": 2614
    },
    {
      "epoch": 2.7182952182952183,
      "grad_norm": 10.824786186218262,
      "learning_rate": 5.0727650727650735e-06,
      "loss": 0.263,
      "step": 2615
    },
    {
      "epoch": 2.7193347193347193,
      "grad_norm": 6.609222412109375,
      "learning_rate": 5.07045507045507e-06,
      "loss": 0.4434,
      "step": 2616
    },
    {
      "epoch": 2.7203742203742203,
      "grad_norm": 4.525443077087402,
      "learning_rate": 5.068145068145068e-06,
      "loss": 0.2959,
      "step": 2617
    },
    {
      "epoch": 2.7214137214137213,
      "grad_norm": 0.9088320732116699,
      "learning_rate": 5.065835065835066e-06,
      "loss": 0.0218,
      "step": 2618
    },
    {
      "epoch": 2.7224532224532223,
      "grad_norm": 11.098006248474121,
      "learning_rate": 5.0635250635250646e-06,
      "loss": 0.5951,
      "step": 2619
    },
    {
      "epoch": 2.7234927234927238,
      "grad_norm": 7.352163314819336,
      "learning_rate": 5.061215061215061e-06,
      "loss": 0.4864,
      "step": 2620
    },
    {
      "epoch": 2.7245322245322248,
      "grad_norm": 0.17772333323955536,
      "learning_rate": 5.058905058905059e-06,
      "loss": 0.0057,
      "step": 2621
    },
    {
      "epoch": 2.725571725571726,
      "grad_norm": 0.08428037911653519,
      "learning_rate": 5.0565950565950574e-06,
      "loss": 0.0021,
      "step": 2622
    },
    {
      "epoch": 2.726611226611227,
      "grad_norm": 0.06792586296796799,
      "learning_rate": 5.054285054285056e-06,
      "loss": 0.0015,
      "step": 2623
    },
    {
      "epoch": 2.727650727650728,
      "grad_norm": 11.568456649780273,
      "learning_rate": 5.051975051975052e-06,
      "loss": 1.3688,
      "step": 2624
    },
    {
      "epoch": 2.728690228690229,
      "grad_norm": 0.07964438199996948,
      "learning_rate": 5.04966504966505e-06,
      "loss": 0.0021,
      "step": 2625
    },
    {
      "epoch": 2.72972972972973,
      "grad_norm": 0.07185760885477066,
      "learning_rate": 5.0473550473550485e-06,
      "loss": 0.0014,
      "step": 2626
    },
    {
      "epoch": 2.730769230769231,
      "grad_norm": 6.794638633728027,
      "learning_rate": 5.045045045045045e-06,
      "loss": 0.221,
      "step": 2627
    },
    {
      "epoch": 2.731808731808732,
      "grad_norm": 7.2844557762146,
      "learning_rate": 5.042735042735043e-06,
      "loss": 0.2326,
      "step": 2628
    },
    {
      "epoch": 2.732848232848233,
      "grad_norm": 2.7436108589172363,
      "learning_rate": 5.040425040425041e-06,
      "loss": 0.0487,
      "step": 2629
    },
    {
      "epoch": 2.733887733887734,
      "grad_norm": 1.4171820878982544,
      "learning_rate": 5.038115038115039e-06,
      "loss": 0.0206,
      "step": 2630
    },
    {
      "epoch": 2.734927234927235,
      "grad_norm": 11.093990325927734,
      "learning_rate": 5.035805035805036e-06,
      "loss": 1.1055,
      "step": 2631
    },
    {
      "epoch": 2.735966735966736,
      "grad_norm": 5.9497599601745605,
      "learning_rate": 5.033495033495034e-06,
      "loss": 0.3682,
      "step": 2632
    },
    {
      "epoch": 2.737006237006237,
      "grad_norm": 5.0965423583984375,
      "learning_rate": 5.031185031185032e-06,
      "loss": 0.1029,
      "step": 2633
    },
    {
      "epoch": 2.738045738045738,
      "grad_norm": 3.5077013969421387,
      "learning_rate": 5.028875028875029e-06,
      "loss": 0.1539,
      "step": 2634
    },
    {
      "epoch": 2.739085239085239,
      "grad_norm": 0.2913844585418701,
      "learning_rate": 5.026565026565027e-06,
      "loss": 0.0082,
      "step": 2635
    },
    {
      "epoch": 2.74012474012474,
      "grad_norm": 5.6061248779296875,
      "learning_rate": 5.0242550242550245e-06,
      "loss": 0.2412,
      "step": 2636
    },
    {
      "epoch": 2.741164241164241,
      "grad_norm": 3.2970173358917236,
      "learning_rate": 5.021945021945023e-06,
      "loss": 0.1059,
      "step": 2637
    },
    {
      "epoch": 2.742203742203742,
      "grad_norm": 4.574480056762695,
      "learning_rate": 5.01963501963502e-06,
      "loss": 0.109,
      "step": 2638
    },
    {
      "epoch": 2.743243243243243,
      "grad_norm": 7.852790355682373,
      "learning_rate": 5.017325017325017e-06,
      "loss": 0.4485,
      "step": 2639
    },
    {
      "epoch": 2.7442827442827444,
      "grad_norm": 1.053780198097229,
      "learning_rate": 5.0150150150150156e-06,
      "loss": 0.0278,
      "step": 2640
    },
    {
      "epoch": 2.7453222453222454,
      "grad_norm": 6.505682468414307,
      "learning_rate": 5.012705012705014e-06,
      "loss": 0.2793,
      "step": 2641
    },
    {
      "epoch": 2.7463617463617465,
      "grad_norm": 9.323391914367676,
      "learning_rate": 5.01039501039501e-06,
      "loss": 0.6098,
      "step": 2642
    },
    {
      "epoch": 2.7474012474012475,
      "grad_norm": 8.641895294189453,
      "learning_rate": 5.0080850080850084e-06,
      "loss": 0.7653,
      "step": 2643
    },
    {
      "epoch": 2.7484407484407485,
      "grad_norm": 9.316137313842773,
      "learning_rate": 5.005775005775007e-06,
      "loss": 0.671,
      "step": 2644
    },
    {
      "epoch": 2.7494802494802495,
      "grad_norm": 2.7795755863189697,
      "learning_rate": 5.003465003465003e-06,
      "loss": 0.0525,
      "step": 2645
    },
    {
      "epoch": 2.7505197505197505,
      "grad_norm": 0.08307638019323349,
      "learning_rate": 5.001155001155001e-06,
      "loss": 0.0021,
      "step": 2646
    },
    {
      "epoch": 2.7515592515592515,
      "grad_norm": 0.11343641579151154,
      "learning_rate": 4.9988449988449995e-06,
      "loss": 0.0022,
      "step": 2647
    },
    {
      "epoch": 2.7525987525987525,
      "grad_norm": 1.186592936515808,
      "learning_rate": 4.996534996534997e-06,
      "loss": 0.0236,
      "step": 2648
    },
    {
      "epoch": 2.7536382536382535,
      "grad_norm": 0.11585807055234909,
      "learning_rate": 4.994224994224995e-06,
      "loss": 0.0026,
      "step": 2649
    },
    {
      "epoch": 2.7546777546777546,
      "grad_norm": 3.3414664268493652,
      "learning_rate": 4.991914991914992e-06,
      "loss": 0.0571,
      "step": 2650
    },
    {
      "epoch": 2.7557172557172556,
      "grad_norm": 4.851938724517822,
      "learning_rate": 4.9896049896049906e-06,
      "loss": 0.3162,
      "step": 2651
    },
    {
      "epoch": 2.756756756756757,
      "grad_norm": 0.75993412733078,
      "learning_rate": 4.987294987294988e-06,
      "loss": 0.0164,
      "step": 2652
    },
    {
      "epoch": 2.757796257796258,
      "grad_norm": 9.393717765808105,
      "learning_rate": 4.984984984984985e-06,
      "loss": 0.3516,
      "step": 2653
    },
    {
      "epoch": 2.758835758835759,
      "grad_norm": 2.778585195541382,
      "learning_rate": 4.9826749826749835e-06,
      "loss": 0.1368,
      "step": 2654
    },
    {
      "epoch": 2.75987525987526,
      "grad_norm": 1.8893013000488281,
      "learning_rate": 4.980364980364981e-06,
      "loss": 0.0591,
      "step": 2655
    },
    {
      "epoch": 2.760914760914761,
      "grad_norm": 0.5797641277313232,
      "learning_rate": 4.978054978054979e-06,
      "loss": 0.0104,
      "step": 2656
    },
    {
      "epoch": 2.761954261954262,
      "grad_norm": 10.824493408203125,
      "learning_rate": 4.975744975744976e-06,
      "loss": 0.3103,
      "step": 2657
    },
    {
      "epoch": 2.762993762993763,
      "grad_norm": 1.135068416595459,
      "learning_rate": 4.973434973434974e-06,
      "loss": 0.029,
      "step": 2658
    },
    {
      "epoch": 2.764033264033264,
      "grad_norm": 7.167160987854004,
      "learning_rate": 4.971124971124972e-06,
      "loss": 0.3448,
      "step": 2659
    },
    {
      "epoch": 2.765072765072765,
      "grad_norm": 0.10508804768323898,
      "learning_rate": 4.968814968814969e-06,
      "loss": 0.0025,
      "step": 2660
    },
    {
      "epoch": 2.766112266112266,
      "grad_norm": 13.092395782470703,
      "learning_rate": 4.9665049665049666e-06,
      "loss": 1.2022,
      "step": 2661
    },
    {
      "epoch": 2.767151767151767,
      "grad_norm": 0.007611572276800871,
      "learning_rate": 4.964194964194965e-06,
      "loss": 0.0002,
      "step": 2662
    },
    {
      "epoch": 2.768191268191268,
      "grad_norm": 2.667224884033203,
      "learning_rate": 4.961884961884962e-06,
      "loss": 0.0571,
      "step": 2663
    },
    {
      "epoch": 2.769230769230769,
      "grad_norm": 2.348841428756714,
      "learning_rate": 4.9595749595749594e-06,
      "loss": 0.0717,
      "step": 2664
    },
    {
      "epoch": 2.77027027027027,
      "grad_norm": 1.9521201848983765,
      "learning_rate": 4.957264957264958e-06,
      "loss": 0.0525,
      "step": 2665
    },
    {
      "epoch": 2.771309771309771,
      "grad_norm": 9.152738571166992,
      "learning_rate": 4.954954954954955e-06,
      "loss": 0.5686,
      "step": 2666
    },
    {
      "epoch": 2.772349272349272,
      "grad_norm": 2.37243390083313,
      "learning_rate": 4.952644952644953e-06,
      "loss": 0.0804,
      "step": 2667
    },
    {
      "epoch": 2.773388773388773,
      "grad_norm": 7.31588888168335,
      "learning_rate": 4.9503349503349505e-06,
      "loss": 0.3661,
      "step": 2668
    },
    {
      "epoch": 2.774428274428274,
      "grad_norm": 5.044492721557617,
      "learning_rate": 4.948024948024949e-06,
      "loss": 0.2454,
      "step": 2669
    },
    {
      "epoch": 2.7754677754677752,
      "grad_norm": 0.23255555331707,
      "learning_rate": 4.945714945714946e-06,
      "loss": 0.0058,
      "step": 2670
    },
    {
      "epoch": 2.7765072765072762,
      "grad_norm": 0.09882773458957672,
      "learning_rate": 4.943404943404943e-06,
      "loss": 0.0019,
      "step": 2671
    },
    {
      "epoch": 2.7775467775467777,
      "grad_norm": 1.9447699785232544,
      "learning_rate": 4.9410949410949416e-06,
      "loss": 0.0621,
      "step": 2672
    },
    {
      "epoch": 2.7785862785862787,
      "grad_norm": 0.6730137467384338,
      "learning_rate": 4.938784938784939e-06,
      "loss": 0.0317,
      "step": 2673
    },
    {
      "epoch": 2.7796257796257797,
      "grad_norm": 9.202969551086426,
      "learning_rate": 4.936474936474937e-06,
      "loss": 0.3289,
      "step": 2674
    },
    {
      "epoch": 2.7806652806652807,
      "grad_norm": 3.387312173843384,
      "learning_rate": 4.9341649341649344e-06,
      "loss": 0.1333,
      "step": 2675
    },
    {
      "epoch": 2.7817047817047817,
      "grad_norm": 0.5006962418556213,
      "learning_rate": 4.931854931854933e-06,
      "loss": 0.0106,
      "step": 2676
    },
    {
      "epoch": 2.7827442827442828,
      "grad_norm": 11.020606994628906,
      "learning_rate": 4.92954492954493e-06,
      "loss": 1.8051,
      "step": 2677
    },
    {
      "epoch": 2.7837837837837838,
      "grad_norm": 11.423202514648438,
      "learning_rate": 4.927234927234928e-06,
      "loss": 0.6795,
      "step": 2678
    },
    {
      "epoch": 2.784823284823285,
      "grad_norm": 4.255502700805664,
      "learning_rate": 4.9249249249249255e-06,
      "loss": 0.334,
      "step": 2679
    },
    {
      "epoch": 2.785862785862786,
      "grad_norm": 3.1610569953918457,
      "learning_rate": 4.922614922614923e-06,
      "loss": 0.1322,
      "step": 2680
    },
    {
      "epoch": 2.786902286902287,
      "grad_norm": 7.8450469970703125,
      "learning_rate": 4.920304920304921e-06,
      "loss": 0.2823,
      "step": 2681
    },
    {
      "epoch": 2.787941787941788,
      "grad_norm": 4.556313991546631,
      "learning_rate": 4.917994917994918e-06,
      "loss": 0.3017,
      "step": 2682
    },
    {
      "epoch": 2.788981288981289,
      "grad_norm": 10.682342529296875,
      "learning_rate": 4.915684915684917e-06,
      "loss": 0.5142,
      "step": 2683
    },
    {
      "epoch": 2.7900207900207903,
      "grad_norm": 1.8365347385406494,
      "learning_rate": 4.913374913374914e-06,
      "loss": 0.0443,
      "step": 2684
    },
    {
      "epoch": 2.7910602910602913,
      "grad_norm": 0.04636334255337715,
      "learning_rate": 4.911064911064911e-06,
      "loss": 0.0008,
      "step": 2685
    },
    {
      "epoch": 2.7920997920997923,
      "grad_norm": 5.025690078735352,
      "learning_rate": 4.9087549087549095e-06,
      "loss": 0.3426,
      "step": 2686
    },
    {
      "epoch": 2.7931392931392933,
      "grad_norm": 8.523011207580566,
      "learning_rate": 4.906444906444907e-06,
      "loss": 0.5284,
      "step": 2687
    },
    {
      "epoch": 2.7941787941787943,
      "grad_norm": 9.12713623046875,
      "learning_rate": 4.904134904134904e-06,
      "loss": 0.393,
      "step": 2688
    },
    {
      "epoch": 2.7952182952182953,
      "grad_norm": 5.500669479370117,
      "learning_rate": 4.901824901824902e-06,
      "loss": 0.1616,
      "step": 2689
    },
    {
      "epoch": 2.7962577962577964,
      "grad_norm": 1.8790087699890137,
      "learning_rate": 4.8995148995149e-06,
      "loss": 0.0492,
      "step": 2690
    },
    {
      "epoch": 2.7972972972972974,
      "grad_norm": 0.5691194534301758,
      "learning_rate": 4.897204897204897e-06,
      "loss": 0.0147,
      "step": 2691
    },
    {
      "epoch": 2.7983367983367984,
      "grad_norm": 12.712238311767578,
      "learning_rate": 4.894894894894895e-06,
      "loss": 1.2213,
      "step": 2692
    },
    {
      "epoch": 2.7993762993762994,
      "grad_norm": 0.6193615794181824,
      "learning_rate": 4.8925848925848926e-06,
      "loss": 0.0165,
      "step": 2693
    },
    {
      "epoch": 2.8004158004158004,
      "grad_norm": 7.124426364898682,
      "learning_rate": 4.890274890274891e-06,
      "loss": 0.4729,
      "step": 2694
    },
    {
      "epoch": 2.8014553014553014,
      "grad_norm": 3.7972962856292725,
      "learning_rate": 4.887964887964888e-06,
      "loss": 0.1297,
      "step": 2695
    },
    {
      "epoch": 2.8024948024948024,
      "grad_norm": 8.690361022949219,
      "learning_rate": 4.885654885654886e-06,
      "loss": 1.0041,
      "step": 2696
    },
    {
      "epoch": 2.8035343035343034,
      "grad_norm": 0.5796958804130554,
      "learning_rate": 4.883344883344884e-06,
      "loss": 0.012,
      "step": 2697
    },
    {
      "epoch": 2.8045738045738045,
      "grad_norm": 5.679058074951172,
      "learning_rate": 4.881034881034881e-06,
      "loss": 0.2276,
      "step": 2698
    },
    {
      "epoch": 2.8056133056133055,
      "grad_norm": 0.24074715375900269,
      "learning_rate": 4.878724878724879e-06,
      "loss": 0.0077,
      "step": 2699
    },
    {
      "epoch": 2.8066528066528065,
      "grad_norm": 0.01518248114734888,
      "learning_rate": 4.8764148764148765e-06,
      "loss": 0.0003,
      "step": 2700
    },
    {
      "epoch": 2.8076923076923075,
      "grad_norm": 10.412557601928711,
      "learning_rate": 4.874104874104875e-06,
      "loss": 0.38,
      "step": 2701
    },
    {
      "epoch": 2.8087318087318085,
      "grad_norm": 0.05676966533064842,
      "learning_rate": 4.871794871794872e-06,
      "loss": 0.0009,
      "step": 2702
    },
    {
      "epoch": 2.8097713097713095,
      "grad_norm": 9.158880233764648,
      "learning_rate": 4.86948486948487e-06,
      "loss": 0.5184,
      "step": 2703
    },
    {
      "epoch": 2.810810810810811,
      "grad_norm": 3.9636144638061523,
      "learning_rate": 4.8671748671748676e-06,
      "loss": 0.1219,
      "step": 2704
    },
    {
      "epoch": 2.811850311850312,
      "grad_norm": 2.21620774269104,
      "learning_rate": 4.864864864864866e-06,
      "loss": 0.0783,
      "step": 2705
    },
    {
      "epoch": 2.812889812889813,
      "grad_norm": 0.4864761233329773,
      "learning_rate": 4.862554862554863e-06,
      "loss": 0.0161,
      "step": 2706
    },
    {
      "epoch": 2.813929313929314,
      "grad_norm": 5.6542158126831055,
      "learning_rate": 4.8602448602448605e-06,
      "loss": 0.4343,
      "step": 2707
    },
    {
      "epoch": 2.814968814968815,
      "grad_norm": 1.9403719902038574,
      "learning_rate": 4.857934857934859e-06,
      "loss": 0.1049,
      "step": 2708
    },
    {
      "epoch": 2.816008316008316,
      "grad_norm": 0.18522211909294128,
      "learning_rate": 4.855624855624856e-06,
      "loss": 0.0049,
      "step": 2709
    },
    {
      "epoch": 2.817047817047817,
      "grad_norm": 0.2546488344669342,
      "learning_rate": 4.853314853314854e-06,
      "loss": 0.0072,
      "step": 2710
    },
    {
      "epoch": 2.818087318087318,
      "grad_norm": 0.18962424993515015,
      "learning_rate": 4.8510048510048515e-06,
      "loss": 0.003,
      "step": 2711
    },
    {
      "epoch": 2.819126819126819,
      "grad_norm": 2.3680453300476074,
      "learning_rate": 4.84869484869485e-06,
      "loss": 0.0469,
      "step": 2712
    },
    {
      "epoch": 2.82016632016632,
      "grad_norm": 0.13189195096492767,
      "learning_rate": 4.846384846384847e-06,
      "loss": 0.0038,
      "step": 2713
    },
    {
      "epoch": 2.821205821205821,
      "grad_norm": 7.136310577392578,
      "learning_rate": 4.844074844074844e-06,
      "loss": 0.2235,
      "step": 2714
    },
    {
      "epoch": 2.822245322245322,
      "grad_norm": 9.05678653717041,
      "learning_rate": 4.841764841764843e-06,
      "loss": 0.7505,
      "step": 2715
    },
    {
      "epoch": 2.8232848232848236,
      "grad_norm": 7.28842306137085,
      "learning_rate": 4.83945483945484e-06,
      "loss": 0.5211,
      "step": 2716
    },
    {
      "epoch": 2.8243243243243246,
      "grad_norm": 5.649138927459717,
      "learning_rate": 4.837144837144837e-06,
      "loss": 0.1875,
      "step": 2717
    },
    {
      "epoch": 2.8253638253638256,
      "grad_norm": 12.433136940002441,
      "learning_rate": 4.8348348348348355e-06,
      "loss": 0.8207,
      "step": 2718
    },
    {
      "epoch": 2.8264033264033266,
      "grad_norm": 8.163975715637207,
      "learning_rate": 4.832524832524833e-06,
      "loss": 0.8473,
      "step": 2719
    },
    {
      "epoch": 2.8274428274428276,
      "grad_norm": 5.121532917022705,
      "learning_rate": 4.83021483021483e-06,
      "loss": 0.1324,
      "step": 2720
    },
    {
      "epoch": 2.8284823284823286,
      "grad_norm": 0.06569239497184753,
      "learning_rate": 4.827904827904828e-06,
      "loss": 0.0021,
      "step": 2721
    },
    {
      "epoch": 2.8295218295218296,
      "grad_norm": 7.412308216094971,
      "learning_rate": 4.825594825594826e-06,
      "loss": 0.7194,
      "step": 2722
    },
    {
      "epoch": 2.8305613305613306,
      "grad_norm": 2.4917805194854736,
      "learning_rate": 4.823284823284824e-06,
      "loss": 0.0472,
      "step": 2723
    },
    {
      "epoch": 2.8316008316008316,
      "grad_norm": 7.543792724609375,
      "learning_rate": 4.820974820974821e-06,
      "loss": 0.5375,
      "step": 2724
    },
    {
      "epoch": 2.8326403326403327,
      "grad_norm": 0.24890974164009094,
      "learning_rate": 4.8186648186648186e-06,
      "loss": 0.0066,
      "step": 2725
    },
    {
      "epoch": 2.8336798336798337,
      "grad_norm": 4.648128032684326,
      "learning_rate": 4.816354816354817e-06,
      "loss": 0.1709,
      "step": 2726
    },
    {
      "epoch": 2.8347193347193347,
      "grad_norm": 5.207144260406494,
      "learning_rate": 4.814044814044814e-06,
      "loss": 0.1652,
      "step": 2727
    },
    {
      "epoch": 2.8357588357588357,
      "grad_norm": 3.2448599338531494,
      "learning_rate": 4.811734811734812e-06,
      "loss": 0.0716,
      "step": 2728
    },
    {
      "epoch": 2.8367983367983367,
      "grad_norm": 4.957944393157959,
      "learning_rate": 4.80942480942481e-06,
      "loss": 0.332,
      "step": 2729
    },
    {
      "epoch": 2.8378378378378377,
      "grad_norm": 0.13377763330936432,
      "learning_rate": 4.807114807114808e-06,
      "loss": 0.0031,
      "step": 2730
    },
    {
      "epoch": 2.8388773388773387,
      "grad_norm": 9.341425895690918,
      "learning_rate": 4.804804804804805e-06,
      "loss": 0.7092,
      "step": 2731
    },
    {
      "epoch": 2.8399168399168397,
      "grad_norm": 4.316988945007324,
      "learning_rate": 4.802494802494803e-06,
      "loss": 0.176,
      "step": 2732
    },
    {
      "epoch": 2.8409563409563408,
      "grad_norm": 0.034980062395334244,
      "learning_rate": 4.800184800184801e-06,
      "loss": 0.0009,
      "step": 2733
    },
    {
      "epoch": 2.8419958419958418,
      "grad_norm": 0.17463620007038116,
      "learning_rate": 4.797874797874798e-06,
      "loss": 0.0038,
      "step": 2734
    },
    {
      "epoch": 2.8430353430353428,
      "grad_norm": 4.154630661010742,
      "learning_rate": 4.795564795564796e-06,
      "loss": 0.1333,
      "step": 2735
    },
    {
      "epoch": 2.8440748440748442,
      "grad_norm": 6.52298641204834,
      "learning_rate": 4.793254793254794e-06,
      "loss": 0.5034,
      "step": 2736
    },
    {
      "epoch": 2.8451143451143452,
      "grad_norm": 4.983566761016846,
      "learning_rate": 4.790944790944792e-06,
      "loss": 0.2822,
      "step": 2737
    },
    {
      "epoch": 2.8461538461538463,
      "grad_norm": 4.204535484313965,
      "learning_rate": 4.788634788634789e-06,
      "loss": 0.2029,
      "step": 2738
    },
    {
      "epoch": 2.8471933471933473,
      "grad_norm": 6.820010662078857,
      "learning_rate": 4.786324786324787e-06,
      "loss": 0.2191,
      "step": 2739
    },
    {
      "epoch": 2.8482328482328483,
      "grad_norm": 3.214674949645996,
      "learning_rate": 4.784014784014785e-06,
      "loss": 0.152,
      "step": 2740
    },
    {
      "epoch": 2.8492723492723493,
      "grad_norm": 7.441554546356201,
      "learning_rate": 4.781704781704782e-06,
      "loss": 0.3018,
      "step": 2741
    },
    {
      "epoch": 2.8503118503118503,
      "grad_norm": 6.902109146118164,
      "learning_rate": 4.77939477939478e-06,
      "loss": 0.4387,
      "step": 2742
    },
    {
      "epoch": 2.8513513513513513,
      "grad_norm": 1.1773970127105713,
      "learning_rate": 4.7770847770847775e-06,
      "loss": 0.0262,
      "step": 2743
    },
    {
      "epoch": 2.8523908523908523,
      "grad_norm": 13.740028381347656,
      "learning_rate": 4.774774774774775e-06,
      "loss": 0.9091,
      "step": 2744
    },
    {
      "epoch": 2.8534303534303533,
      "grad_norm": 2.1474075317382812,
      "learning_rate": 4.772464772464773e-06,
      "loss": 0.0869,
      "step": 2745
    },
    {
      "epoch": 2.8544698544698544,
      "grad_norm": 5.147608757019043,
      "learning_rate": 4.77015477015477e-06,
      "loss": 0.4671,
      "step": 2746
    },
    {
      "epoch": 2.8555093555093554,
      "grad_norm": 0.08837344497442245,
      "learning_rate": 4.767844767844768e-06,
      "loss": 0.0027,
      "step": 2747
    },
    {
      "epoch": 2.856548856548857,
      "grad_norm": 9.307243347167969,
      "learning_rate": 4.765534765534766e-06,
      "loss": 0.2708,
      "step": 2748
    },
    {
      "epoch": 2.857588357588358,
      "grad_norm": 0.17253902554512024,
      "learning_rate": 4.763224763224763e-06,
      "loss": 0.003,
      "step": 2749
    },
    {
      "epoch": 2.858627858627859,
      "grad_norm": 0.9690131545066833,
      "learning_rate": 4.7609147609147615e-06,
      "loss": 0.0471,
      "step": 2750
    },
    {
      "epoch": 2.85966735966736,
      "grad_norm": 10.780189514160156,
      "learning_rate": 4.758604758604759e-06,
      "loss": 0.7136,
      "step": 2751
    },
    {
      "epoch": 2.860706860706861,
      "grad_norm": 0.4083833396434784,
      "learning_rate": 4.756294756294756e-06,
      "loss": 0.0103,
      "step": 2752
    },
    {
      "epoch": 2.861746361746362,
      "grad_norm": 7.011819362640381,
      "learning_rate": 4.753984753984754e-06,
      "loss": 0.1977,
      "step": 2753
    },
    {
      "epoch": 2.862785862785863,
      "grad_norm": 0.6745598316192627,
      "learning_rate": 4.751674751674752e-06,
      "loss": 0.0166,
      "step": 2754
    },
    {
      "epoch": 2.863825363825364,
      "grad_norm": 2.218971014022827,
      "learning_rate": 4.74936474936475e-06,
      "loss": 0.0656,
      "step": 2755
    },
    {
      "epoch": 2.864864864864865,
      "grad_norm": 0.19676561653614044,
      "learning_rate": 4.747054747054747e-06,
      "loss": 0.0069,
      "step": 2756
    },
    {
      "epoch": 2.865904365904366,
      "grad_norm": 8.756501197814941,
      "learning_rate": 4.7447447447447454e-06,
      "loss": 0.972,
      "step": 2757
    },
    {
      "epoch": 2.866943866943867,
      "grad_norm": 0.0635393038392067,
      "learning_rate": 4.742434742434743e-06,
      "loss": 0.0023,
      "step": 2758
    },
    {
      "epoch": 2.867983367983368,
      "grad_norm": 6.238755226135254,
      "learning_rate": 4.740124740124741e-06,
      "loss": 0.2034,
      "step": 2759
    },
    {
      "epoch": 2.869022869022869,
      "grad_norm": 8.922481536865234,
      "learning_rate": 4.737814737814738e-06,
      "loss": 0.6653,
      "step": 2760
    },
    {
      "epoch": 2.87006237006237,
      "grad_norm": 0.8628990650177002,
      "learning_rate": 4.735504735504736e-06,
      "loss": 0.0221,
      "step": 2761
    },
    {
      "epoch": 2.871101871101871,
      "grad_norm": 0.3078502118587494,
      "learning_rate": 4.733194733194734e-06,
      "loss": 0.0061,
      "step": 2762
    },
    {
      "epoch": 2.872141372141372,
      "grad_norm": 13.62101936340332,
      "learning_rate": 4.730884730884731e-06,
      "loss": 0.8565,
      "step": 2763
    },
    {
      "epoch": 2.873180873180873,
      "grad_norm": 3.0611491203308105,
      "learning_rate": 4.728574728574729e-06,
      "loss": 0.0806,
      "step": 2764
    },
    {
      "epoch": 2.874220374220374,
      "grad_norm": 0.7029672861099243,
      "learning_rate": 4.726264726264727e-06,
      "loss": 0.0181,
      "step": 2765
    },
    {
      "epoch": 2.875259875259875,
      "grad_norm": 0.2899930775165558,
      "learning_rate": 4.723954723954725e-06,
      "loss": 0.0055,
      "step": 2766
    },
    {
      "epoch": 2.876299376299376,
      "grad_norm": 1.9155933856964111,
      "learning_rate": 4.721644721644722e-06,
      "loss": 0.0693,
      "step": 2767
    },
    {
      "epoch": 2.8773388773388775,
      "grad_norm": 8.559905052185059,
      "learning_rate": 4.71933471933472e-06,
      "loss": 0.3092,
      "step": 2768
    },
    {
      "epoch": 2.8783783783783785,
      "grad_norm": 0.37766215205192566,
      "learning_rate": 4.717024717024718e-06,
      "loss": 0.0097,
      "step": 2769
    },
    {
      "epoch": 2.8794178794178795,
      "grad_norm": 0.4770283102989197,
      "learning_rate": 4.714714714714715e-06,
      "loss": 0.012,
      "step": 2770
    },
    {
      "epoch": 2.8804573804573805,
      "grad_norm": 0.04901864379644394,
      "learning_rate": 4.7124047124047125e-06,
      "loss": 0.0016,
      "step": 2771
    },
    {
      "epoch": 2.8814968814968815,
      "grad_norm": 10.537938117980957,
      "learning_rate": 4.710094710094711e-06,
      "loss": 0.7172,
      "step": 2772
    },
    {
      "epoch": 2.8825363825363826,
      "grad_norm": 1.2608181238174438,
      "learning_rate": 4.707784707784708e-06,
      "loss": 0.0336,
      "step": 2773
    },
    {
      "epoch": 2.8835758835758836,
      "grad_norm": 0.531566321849823,
      "learning_rate": 4.705474705474705e-06,
      "loss": 0.0206,
      "step": 2774
    },
    {
      "epoch": 2.8846153846153846,
      "grad_norm": 5.012516498565674,
      "learning_rate": 4.7031647031647035e-06,
      "loss": 0.2701,
      "step": 2775
    },
    {
      "epoch": 2.8856548856548856,
      "grad_norm": 2.223706007003784,
      "learning_rate": 4.700854700854701e-06,
      "loss": 0.0858,
      "step": 2776
    },
    {
      "epoch": 2.8866943866943866,
      "grad_norm": 2.592135190963745,
      "learning_rate": 4.698544698544699e-06,
      "loss": 0.073,
      "step": 2777
    },
    {
      "epoch": 2.8877338877338876,
      "grad_norm": 11.445849418640137,
      "learning_rate": 4.696234696234696e-06,
      "loss": 1.3401,
      "step": 2778
    },
    {
      "epoch": 2.8887733887733886,
      "grad_norm": 1.9957225322723389,
      "learning_rate": 4.693924693924694e-06,
      "loss": 0.0379,
      "step": 2779
    },
    {
      "epoch": 2.88981288981289,
      "grad_norm": 0.9727466702461243,
      "learning_rate": 4.691614691614692e-06,
      "loss": 0.0179,
      "step": 2780
    },
    {
      "epoch": 2.890852390852391,
      "grad_norm": 2.8076136112213135,
      "learning_rate": 4.689304689304689e-06,
      "loss": 0.0701,
      "step": 2781
    },
    {
      "epoch": 2.891891891891892,
      "grad_norm": 0.028624217957258224,
      "learning_rate": 4.6869946869946875e-06,
      "loss": 0.0006,
      "step": 2782
    },
    {
      "epoch": 2.892931392931393,
      "grad_norm": 2.858949899673462,
      "learning_rate": 4.684684684684685e-06,
      "loss": 0.1371,
      "step": 2783
    },
    {
      "epoch": 2.893970893970894,
      "grad_norm": 1.272584319114685,
      "learning_rate": 4.682374682374683e-06,
      "loss": 0.0191,
      "step": 2784
    },
    {
      "epoch": 2.895010395010395,
      "grad_norm": 2.8619091510772705,
      "learning_rate": 4.68006468006468e-06,
      "loss": 0.1527,
      "step": 2785
    },
    {
      "epoch": 2.896049896049896,
      "grad_norm": 3.4362850189208984,
      "learning_rate": 4.6777546777546786e-06,
      "loss": 0.3499,
      "step": 2786
    },
    {
      "epoch": 2.897089397089397,
      "grad_norm": 5.900140762329102,
      "learning_rate": 4.675444675444676e-06,
      "loss": 0.2825,
      "step": 2787
    },
    {
      "epoch": 2.898128898128898,
      "grad_norm": 12.06615161895752,
      "learning_rate": 4.673134673134673e-06,
      "loss": 1.3305,
      "step": 2788
    },
    {
      "epoch": 2.899168399168399,
      "grad_norm": 8.633148193359375,
      "learning_rate": 4.6708246708246714e-06,
      "loss": 0.3222,
      "step": 2789
    },
    {
      "epoch": 2.9002079002079,
      "grad_norm": 7.318634986877441,
      "learning_rate": 4.668514668514669e-06,
      "loss": 0.2946,
      "step": 2790
    },
    {
      "epoch": 2.901247401247401,
      "grad_norm": 0.35540270805358887,
      "learning_rate": 4.666204666204667e-06,
      "loss": 0.0078,
      "step": 2791
    },
    {
      "epoch": 2.9022869022869022,
      "grad_norm": 0.6462912559509277,
      "learning_rate": 4.663894663894664e-06,
      "loss": 0.0091,
      "step": 2792
    },
    {
      "epoch": 2.9033264033264032,
      "grad_norm": 0.5021451711654663,
      "learning_rate": 4.6615846615846625e-06,
      "loss": 0.0135,
      "step": 2793
    },
    {
      "epoch": 2.9043659043659042,
      "grad_norm": 0.36885151267051697,
      "learning_rate": 4.65927465927466e-06,
      "loss": 0.0108,
      "step": 2794
    },
    {
      "epoch": 2.9054054054054053,
      "grad_norm": 5.507650375366211,
      "learning_rate": 4.656964656964657e-06,
      "loss": 0.329,
      "step": 2795
    },
    {
      "epoch": 2.9064449064449063,
      "grad_norm": 0.02889976091682911,
      "learning_rate": 4.654654654654655e-06,
      "loss": 0.0008,
      "step": 2796
    },
    {
      "epoch": 2.9074844074844073,
      "grad_norm": 4.460318565368652,
      "learning_rate": 4.652344652344653e-06,
      "loss": 0.2253,
      "step": 2797
    },
    {
      "epoch": 2.9085239085239083,
      "grad_norm": 0.10087776929140091,
      "learning_rate": 4.65003465003465e-06,
      "loss": 0.0025,
      "step": 2798
    },
    {
      "epoch": 2.9095634095634093,
      "grad_norm": 0.39357542991638184,
      "learning_rate": 4.647724647724648e-06,
      "loss": 0.0094,
      "step": 2799
    },
    {
      "epoch": 2.9106029106029108,
      "grad_norm": 0.20674152672290802,
      "learning_rate": 4.645414645414646e-06,
      "loss": 0.0045,
      "step": 2800
    },
    {
      "epoch": 2.9116424116424118,
      "grad_norm": 1.4034574031829834,
      "learning_rate": 4.643104643104644e-06,
      "loss": 0.0323,
      "step": 2801
    },
    {
      "epoch": 2.912681912681913,
      "grad_norm": 7.4032416343688965,
      "learning_rate": 4.640794640794641e-06,
      "loss": 0.5195,
      "step": 2802
    },
    {
      "epoch": 2.913721413721414,
      "grad_norm": 3.127525806427002,
      "learning_rate": 4.6384846384846385e-06,
      "loss": 0.0481,
      "step": 2803
    },
    {
      "epoch": 2.914760914760915,
      "grad_norm": 1.7245556116104126,
      "learning_rate": 4.636174636174637e-06,
      "loss": 0.0282,
      "step": 2804
    },
    {
      "epoch": 2.915800415800416,
      "grad_norm": 9.243249893188477,
      "learning_rate": 4.633864633864634e-06,
      "loss": 0.3518,
      "step": 2805
    },
    {
      "epoch": 2.916839916839917,
      "grad_norm": 2.03916597366333,
      "learning_rate": 4.631554631554631e-06,
      "loss": 0.0527,
      "step": 2806
    },
    {
      "epoch": 2.917879417879418,
      "grad_norm": 7.111269474029541,
      "learning_rate": 4.6292446292446295e-06,
      "loss": 0.2616,
      "step": 2807
    },
    {
      "epoch": 2.918918918918919,
      "grad_norm": 3.627591848373413,
      "learning_rate": 4.626934626934627e-06,
      "loss": 0.088,
      "step": 2808
    },
    {
      "epoch": 2.91995841995842,
      "grad_norm": 2.4315571784973145,
      "learning_rate": 4.624624624624625e-06,
      "loss": 0.0711,
      "step": 2809
    },
    {
      "epoch": 2.920997920997921,
      "grad_norm": 7.548877716064453,
      "learning_rate": 4.6223146223146224e-06,
      "loss": 0.6083,
      "step": 2810
    },
    {
      "epoch": 2.922037422037422,
      "grad_norm": 13.10161304473877,
      "learning_rate": 4.620004620004621e-06,
      "loss": 1.686,
      "step": 2811
    },
    {
      "epoch": 2.9230769230769234,
      "grad_norm": 1.7761461734771729,
      "learning_rate": 4.617694617694618e-06,
      "loss": 0.036,
      "step": 2812
    },
    {
      "epoch": 2.9241164241164244,
      "grad_norm": 0.5098281502723694,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.0071,
      "step": 2813
    },
    {
      "epoch": 2.9251559251559254,
      "grad_norm": 337.7049255371094,
      "learning_rate": 4.6130746130746135e-06,
      "loss": 0.4585,
      "step": 2814
    },
    {
      "epoch": 2.9261954261954264,
      "grad_norm": 4.641257286071777,
      "learning_rate": 4.610764610764611e-06,
      "loss": 0.1252,
      "step": 2815
    },
    {
      "epoch": 2.9272349272349274,
      "grad_norm": 2.1013717651367188,
      "learning_rate": 4.608454608454609e-06,
      "loss": 0.0614,
      "step": 2816
    },
    {
      "epoch": 2.9282744282744284,
      "grad_norm": 2.736609697341919,
      "learning_rate": 4.606144606144606e-06,
      "loss": 0.0907,
      "step": 2817
    },
    {
      "epoch": 2.9293139293139294,
      "grad_norm": 8.260156631469727,
      "learning_rate": 4.6038346038346046e-06,
      "loss": 0.4214,
      "step": 2818
    },
    {
      "epoch": 2.9303534303534304,
      "grad_norm": 0.2285541146993637,
      "learning_rate": 4.601524601524602e-06,
      "loss": 0.0052,
      "step": 2819
    },
    {
      "epoch": 2.9313929313929314,
      "grad_norm": 2.672311782836914,
      "learning_rate": 4.5992145992146e-06,
      "loss": 0.0733,
      "step": 2820
    },
    {
      "epoch": 2.9324324324324325,
      "grad_norm": 3.377425193786621,
      "learning_rate": 4.5969045969045974e-06,
      "loss": 0.0867,
      "step": 2821
    },
    {
      "epoch": 2.9334719334719335,
      "grad_norm": 9.351137161254883,
      "learning_rate": 4.594594594594596e-06,
      "loss": 0.7431,
      "step": 2822
    },
    {
      "epoch": 2.9345114345114345,
      "grad_norm": 0.11065870523452759,
      "learning_rate": 4.592284592284593e-06,
      "loss": 0.002,
      "step": 2823
    },
    {
      "epoch": 2.9355509355509355,
      "grad_norm": 0.9388895630836487,
      "learning_rate": 4.58997458997459e-06,
      "loss": 0.0377,
      "step": 2824
    },
    {
      "epoch": 2.9365904365904365,
      "grad_norm": 1.9601761102676392,
      "learning_rate": 4.5876645876645885e-06,
      "loss": 0.0892,
      "step": 2825
    },
    {
      "epoch": 2.9376299376299375,
      "grad_norm": 5.889634132385254,
      "learning_rate": 4.585354585354586e-06,
      "loss": 0.3781,
      "step": 2826
    },
    {
      "epoch": 2.9386694386694385,
      "grad_norm": 5.7516770362854,
      "learning_rate": 4.583044583044583e-06,
      "loss": 0.4462,
      "step": 2827
    },
    {
      "epoch": 2.9397089397089395,
      "grad_norm": 2.089272975921631,
      "learning_rate": 4.580734580734581e-06,
      "loss": 0.0607,
      "step": 2828
    },
    {
      "epoch": 2.9407484407484406,
      "grad_norm": 10.779491424560547,
      "learning_rate": 4.578424578424579e-06,
      "loss": 0.3636,
      "step": 2829
    },
    {
      "epoch": 2.9417879417879416,
      "grad_norm": 2.810117721557617,
      "learning_rate": 4.576114576114576e-06,
      "loss": 0.0683,
      "step": 2830
    },
    {
      "epoch": 2.9428274428274426,
      "grad_norm": 0.02683236077427864,
      "learning_rate": 4.573804573804574e-06,
      "loss": 0.0007,
      "step": 2831
    },
    {
      "epoch": 2.943866943866944,
      "grad_norm": 0.10576693713665009,
      "learning_rate": 4.571494571494572e-06,
      "loss": 0.0034,
      "step": 2832
    },
    {
      "epoch": 2.944906444906445,
      "grad_norm": 1.2334619760513306,
      "learning_rate": 4.569184569184569e-06,
      "loss": 0.0306,
      "step": 2833
    },
    {
      "epoch": 2.945945945945946,
      "grad_norm": 8.295530319213867,
      "learning_rate": 4.566874566874567e-06,
      "loss": 0.359,
      "step": 2834
    },
    {
      "epoch": 2.946985446985447,
      "grad_norm": 3.9480338096618652,
      "learning_rate": 4.5645645645645645e-06,
      "loss": 0.0801,
      "step": 2835
    },
    {
      "epoch": 2.948024948024948,
      "grad_norm": 2.964906692504883,
      "learning_rate": 4.562254562254563e-06,
      "loss": 0.0966,
      "step": 2836
    },
    {
      "epoch": 2.949064449064449,
      "grad_norm": 9.53171443939209,
      "learning_rate": 4.55994455994456e-06,
      "loss": 0.694,
      "step": 2837
    },
    {
      "epoch": 2.95010395010395,
      "grad_norm": 0.06429767608642578,
      "learning_rate": 4.557634557634558e-06,
      "loss": 0.0014,
      "step": 2838
    },
    {
      "epoch": 2.951143451143451,
      "grad_norm": 1.9759715795516968,
      "learning_rate": 4.5553245553245556e-06,
      "loss": 0.0314,
      "step": 2839
    },
    {
      "epoch": 2.952182952182952,
      "grad_norm": 5.269041538238525,
      "learning_rate": 4.553014553014554e-06,
      "loss": 0.1935,
      "step": 2840
    },
    {
      "epoch": 2.953222453222453,
      "grad_norm": 5.443349361419678,
      "learning_rate": 4.550704550704551e-06,
      "loss": 0.1412,
      "step": 2841
    },
    {
      "epoch": 2.954261954261954,
      "grad_norm": 11.945755004882812,
      "learning_rate": 4.5483945483945484e-06,
      "loss": 1.0154,
      "step": 2842
    },
    {
      "epoch": 2.955301455301455,
      "grad_norm": 2.846902847290039,
      "learning_rate": 4.546084546084547e-06,
      "loss": 0.0632,
      "step": 2843
    },
    {
      "epoch": 2.9563409563409566,
      "grad_norm": 10.10546588897705,
      "learning_rate": 4.543774543774544e-06,
      "loss": 0.4177,
      "step": 2844
    },
    {
      "epoch": 2.9573804573804576,
      "grad_norm": 0.19464100897312164,
      "learning_rate": 4.541464541464542e-06,
      "loss": 0.0054,
      "step": 2845
    },
    {
      "epoch": 2.9584199584199586,
      "grad_norm": 7.272435665130615,
      "learning_rate": 4.5391545391545395e-06,
      "loss": 0.3266,
      "step": 2846
    },
    {
      "epoch": 2.9594594594594597,
      "grad_norm": 0.25439026951789856,
      "learning_rate": 4.536844536844538e-06,
      "loss": 0.0033,
      "step": 2847
    },
    {
      "epoch": 2.9604989604989607,
      "grad_norm": 0.2873782217502594,
      "learning_rate": 4.534534534534535e-06,
      "loss": 0.0045,
      "step": 2848
    },
    {
      "epoch": 2.9615384615384617,
      "grad_norm": 0.18630096316337585,
      "learning_rate": 4.532224532224533e-06,
      "loss": 0.0037,
      "step": 2849
    },
    {
      "epoch": 2.9625779625779627,
      "grad_norm": 1.1513333320617676,
      "learning_rate": 4.5299145299145306e-06,
      "loss": 0.0216,
      "step": 2850
    },
    {
      "epoch": 2.9636174636174637,
      "grad_norm": 6.185680866241455,
      "learning_rate": 4.527604527604528e-06,
      "loss": 0.2914,
      "step": 2851
    },
    {
      "epoch": 2.9646569646569647,
      "grad_norm": 2.607320785522461,
      "learning_rate": 4.525294525294526e-06,
      "loss": 0.2749,
      "step": 2852
    },
    {
      "epoch": 2.9656964656964657,
      "grad_norm": 15.155496597290039,
      "learning_rate": 4.5229845229845235e-06,
      "loss": 1.5991,
      "step": 2853
    },
    {
      "epoch": 2.9667359667359667,
      "grad_norm": 1.463409185409546,
      "learning_rate": 4.520674520674521e-06,
      "loss": 0.0406,
      "step": 2854
    },
    {
      "epoch": 2.9677754677754677,
      "grad_norm": 1.1975452899932861,
      "learning_rate": 4.518364518364519e-06,
      "loss": 0.0287,
      "step": 2855
    },
    {
      "epoch": 2.9688149688149688,
      "grad_norm": 0.2848760187625885,
      "learning_rate": 4.516054516054516e-06,
      "loss": 0.0038,
      "step": 2856
    },
    {
      "epoch": 2.9698544698544698,
      "grad_norm": 14.927924156188965,
      "learning_rate": 4.513744513744514e-06,
      "loss": 1.2295,
      "step": 2857
    },
    {
      "epoch": 2.970893970893971,
      "grad_norm": 5.528604507446289,
      "learning_rate": 4.511434511434512e-06,
      "loss": 0.2973,
      "step": 2858
    },
    {
      "epoch": 2.971933471933472,
      "grad_norm": 5.287175178527832,
      "learning_rate": 4.509124509124509e-06,
      "loss": 0.1407,
      "step": 2859
    },
    {
      "epoch": 2.972972972972973,
      "grad_norm": 2.9800124168395996,
      "learning_rate": 4.5068145068145066e-06,
      "loss": 0.0748,
      "step": 2860
    },
    {
      "epoch": 2.974012474012474,
      "grad_norm": 3.8337507247924805,
      "learning_rate": 4.504504504504505e-06,
      "loss": 0.1819,
      "step": 2861
    },
    {
      "epoch": 2.975051975051975,
      "grad_norm": 8.198847770690918,
      "learning_rate": 4.502194502194502e-06,
      "loss": 0.2563,
      "step": 2862
    },
    {
      "epoch": 2.976091476091476,
      "grad_norm": 4.668789863586426,
      "learning_rate": 4.4998844998845e-06,
      "loss": 0.1607,
      "step": 2863
    },
    {
      "epoch": 2.9771309771309773,
      "grad_norm": 3.7583305835723877,
      "learning_rate": 4.497574497574498e-06,
      "loss": 0.1952,
      "step": 2864
    },
    {
      "epoch": 2.9781704781704783,
      "grad_norm": 1.488911747932434,
      "learning_rate": 4.495264495264496e-06,
      "loss": 0.0382,
      "step": 2865
    },
    {
      "epoch": 2.9792099792099793,
      "grad_norm": 0.9805212020874023,
      "learning_rate": 4.492954492954493e-06,
      "loss": 0.2726,
      "step": 2866
    },
    {
      "epoch": 2.9802494802494803,
      "grad_norm": 0.03673991560935974,
      "learning_rate": 4.490644490644491e-06,
      "loss": 0.0005,
      "step": 2867
    },
    {
      "epoch": 2.9812889812889813,
      "grad_norm": 1.645326018333435,
      "learning_rate": 4.488334488334489e-06,
      "loss": 0.0279,
      "step": 2868
    },
    {
      "epoch": 2.9823284823284824,
      "grad_norm": 2.383275032043457,
      "learning_rate": 4.486024486024486e-06,
      "loss": 0.117,
      "step": 2869
    },
    {
      "epoch": 2.9833679833679834,
      "grad_norm": 1.2870901823043823,
      "learning_rate": 4.483714483714484e-06,
      "loss": 0.0229,
      "step": 2870
    },
    {
      "epoch": 2.9844074844074844,
      "grad_norm": 0.06789594888687134,
      "learning_rate": 4.4814044814044816e-06,
      "loss": 0.0017,
      "step": 2871
    },
    {
      "epoch": 2.9854469854469854,
      "grad_norm": 0.3903137445449829,
      "learning_rate": 4.47909447909448e-06,
      "loss": 0.0081,
      "step": 2872
    },
    {
      "epoch": 2.9864864864864864,
      "grad_norm": 9.606902122497559,
      "learning_rate": 4.476784476784477e-06,
      "loss": 1.0147,
      "step": 2873
    },
    {
      "epoch": 2.9875259875259874,
      "grad_norm": 4.721467971801758,
      "learning_rate": 4.474474474474475e-06,
      "loss": 0.0904,
      "step": 2874
    },
    {
      "epoch": 2.9885654885654884,
      "grad_norm": 0.037113163620233536,
      "learning_rate": 4.472164472164473e-06,
      "loss": 0.0009,
      "step": 2875
    },
    {
      "epoch": 2.98960498960499,
      "grad_norm": 3.5482118129730225,
      "learning_rate": 4.469854469854471e-06,
      "loss": 0.1736,
      "step": 2876
    },
    {
      "epoch": 2.990644490644491,
      "grad_norm": 1.804797887802124,
      "learning_rate": 4.467544467544468e-06,
      "loss": 0.0349,
      "step": 2877
    },
    {
      "epoch": 2.991683991683992,
      "grad_norm": 0.5807712078094482,
      "learning_rate": 4.4652344652344655e-06,
      "loss": 0.0149,
      "step": 2878
    },
    {
      "epoch": 2.992723492723493,
      "grad_norm": 8.496133804321289,
      "learning_rate": 4.462924462924464e-06,
      "loss": 0.2932,
      "step": 2879
    },
    {
      "epoch": 2.993762993762994,
      "grad_norm": 4.594316482543945,
      "learning_rate": 4.460614460614461e-06,
      "loss": 0.1771,
      "step": 2880
    },
    {
      "epoch": 2.994802494802495,
      "grad_norm": 1.7351125478744507,
      "learning_rate": 4.458304458304458e-06,
      "loss": 0.0422,
      "step": 2881
    },
    {
      "epoch": 2.995841995841996,
      "grad_norm": 14.885557174682617,
      "learning_rate": 4.455994455994457e-06,
      "loss": 1.0164,
      "step": 2882
    },
    {
      "epoch": 2.996881496881497,
      "grad_norm": 0.8425683975219727,
      "learning_rate": 4.453684453684454e-06,
      "loss": 0.0199,
      "step": 2883
    },
    {
      "epoch": 2.997920997920998,
      "grad_norm": 0.0038593353237956762,
      "learning_rate": 4.451374451374451e-06,
      "loss": 0.0001,
      "step": 2884
    },
    {
      "epoch": 2.998960498960499,
      "grad_norm": 4.419792175292969,
      "learning_rate": 4.4490644490644495e-06,
      "loss": 0.1397,
      "step": 2885
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.9082928895950317,
      "learning_rate": 4.446754446754447e-06,
      "loss": 0.15,
      "step": 2886
    },
    {
      "epoch": 3.001039501039501,
      "grad_norm": 0.09658217430114746,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.0024,
      "step": 2887
    },
    {
      "epoch": 3.002079002079002,
      "grad_norm": 0.6714063882827759,
      "learning_rate": 4.442134442134442e-06,
      "loss": 0.0136,
      "step": 2888
    },
    {
      "epoch": 3.003118503118503,
      "grad_norm": 0.06490445137023926,
      "learning_rate": 4.43982443982444e-06,
      "loss": 0.0019,
      "step": 2889
    },
    {
      "epoch": 3.004158004158004,
      "grad_norm": 3.7229204177856445,
      "learning_rate": 4.437514437514438e-06,
      "loss": 0.256,
      "step": 2890
    },
    {
      "epoch": 3.005197505197505,
      "grad_norm": 1.2547167539596558,
      "learning_rate": 4.435204435204435e-06,
      "loss": 0.0268,
      "step": 2891
    },
    {
      "epoch": 3.006237006237006,
      "grad_norm": 0.8360212445259094,
      "learning_rate": 4.432894432894433e-06,
      "loss": 0.0228,
      "step": 2892
    },
    {
      "epoch": 3.007276507276507,
      "grad_norm": 9.071552276611328,
      "learning_rate": 4.430584430584431e-06,
      "loss": 0.2006,
      "step": 2893
    },
    {
      "epoch": 3.008316008316008,
      "grad_norm": 3.2643280029296875,
      "learning_rate": 4.428274428274429e-06,
      "loss": 0.1102,
      "step": 2894
    },
    {
      "epoch": 3.0093555093555096,
      "grad_norm": 8.89312744140625,
      "learning_rate": 4.425964425964426e-06,
      "loss": 0.9518,
      "step": 2895
    },
    {
      "epoch": 3.0103950103950106,
      "grad_norm": 1.1194871664047241,
      "learning_rate": 4.423654423654424e-06,
      "loss": 0.0156,
      "step": 2896
    },
    {
      "epoch": 3.0114345114345116,
      "grad_norm": 0.031017301604151726,
      "learning_rate": 4.421344421344422e-06,
      "loss": 0.0007,
      "step": 2897
    },
    {
      "epoch": 3.0124740124740126,
      "grad_norm": 9.319890022277832,
      "learning_rate": 4.419034419034419e-06,
      "loss": 0.2443,
      "step": 2898
    },
    {
      "epoch": 3.0135135135135136,
      "grad_norm": 0.7562645077705383,
      "learning_rate": 4.416724416724417e-06,
      "loss": 0.0138,
      "step": 2899
    },
    {
      "epoch": 3.0145530145530146,
      "grad_norm": 5.006690502166748,
      "learning_rate": 4.414414414414415e-06,
      "loss": 0.0781,
      "step": 2900
    },
    {
      "epoch": 3.0155925155925156,
      "grad_norm": 1.8787916898727417,
      "learning_rate": 4.412104412104413e-06,
      "loss": 0.0643,
      "step": 2901
    },
    {
      "epoch": 3.0166320166320166,
      "grad_norm": 7.709594249725342,
      "learning_rate": 4.40979440979441e-06,
      "loss": 0.5319,
      "step": 2902
    },
    {
      "epoch": 3.0176715176715176,
      "grad_norm": 8.625652313232422,
      "learning_rate": 4.4074844074844084e-06,
      "loss": 0.214,
      "step": 2903
    },
    {
      "epoch": 3.0187110187110187,
      "grad_norm": 8.986515045166016,
      "learning_rate": 4.405174405174406e-06,
      "loss": 0.4628,
      "step": 2904
    },
    {
      "epoch": 3.0197505197505197,
      "grad_norm": 6.721210956573486,
      "learning_rate": 4.402864402864403e-06,
      "loss": 0.269,
      "step": 2905
    },
    {
      "epoch": 3.0207900207900207,
      "grad_norm": 9.072442054748535,
      "learning_rate": 4.400554400554401e-06,
      "loss": 0.3887,
      "step": 2906
    },
    {
      "epoch": 3.0218295218295217,
      "grad_norm": 0.03076879307627678,
      "learning_rate": 4.398244398244399e-06,
      "loss": 0.0007,
      "step": 2907
    },
    {
      "epoch": 3.0228690228690227,
      "grad_norm": 1.009333848953247,
      "learning_rate": 4.395934395934397e-06,
      "loss": 0.2635,
      "step": 2908
    },
    {
      "epoch": 3.0239085239085237,
      "grad_norm": 2.1226351261138916,
      "learning_rate": 4.393624393624394e-06,
      "loss": 0.0672,
      "step": 2909
    },
    {
      "epoch": 3.024948024948025,
      "grad_norm": 8.85048770904541,
      "learning_rate": 4.3913143913143915e-06,
      "loss": 0.7063,
      "step": 2910
    },
    {
      "epoch": 3.025987525987526,
      "grad_norm": 2.2179903984069824,
      "learning_rate": 4.38900438900439e-06,
      "loss": 0.1219,
      "step": 2911
    },
    {
      "epoch": 3.027027027027027,
      "grad_norm": 2.5947070121765137,
      "learning_rate": 4.386694386694387e-06,
      "loss": 0.0804,
      "step": 2912
    },
    {
      "epoch": 3.028066528066528,
      "grad_norm": 6.031675815582275,
      "learning_rate": 4.384384384384384e-06,
      "loss": 0.1997,
      "step": 2913
    },
    {
      "epoch": 3.029106029106029,
      "grad_norm": 1.8736892938613892,
      "learning_rate": 4.382074382074383e-06,
      "loss": 0.3105,
      "step": 2914
    },
    {
      "epoch": 3.0301455301455302,
      "grad_norm": 0.2389967292547226,
      "learning_rate": 4.37976437976438e-06,
      "loss": 0.0042,
      "step": 2915
    },
    {
      "epoch": 3.0311850311850312,
      "grad_norm": 0.41082292795181274,
      "learning_rate": 4.377454377454377e-06,
      "loss": 0.008,
      "step": 2916
    },
    {
      "epoch": 3.0322245322245323,
      "grad_norm": 5.320422172546387,
      "learning_rate": 4.3751443751443755e-06,
      "loss": 0.1933,
      "step": 2917
    },
    {
      "epoch": 3.0332640332640333,
      "grad_norm": 3.031292676925659,
      "learning_rate": 4.372834372834373e-06,
      "loss": 0.0862,
      "step": 2918
    },
    {
      "epoch": 3.0343035343035343,
      "grad_norm": 4.29350471496582,
      "learning_rate": 4.370524370524371e-06,
      "loss": 0.3082,
      "step": 2919
    },
    {
      "epoch": 3.0353430353430353,
      "grad_norm": 9.461339950561523,
      "learning_rate": 4.368214368214368e-06,
      "loss": 0.4198,
      "step": 2920
    },
    {
      "epoch": 3.0363825363825363,
      "grad_norm": 2.768932580947876,
      "learning_rate": 4.3659043659043665e-06,
      "loss": 0.0551,
      "step": 2921
    },
    {
      "epoch": 3.0374220374220373,
      "grad_norm": 18.20948600769043,
      "learning_rate": 4.363594363594364e-06,
      "loss": 0.7916,
      "step": 2922
    },
    {
      "epoch": 3.0384615384615383,
      "grad_norm": 11.076471328735352,
      "learning_rate": 4.361284361284361e-06,
      "loss": 0.8374,
      "step": 2923
    },
    {
      "epoch": 3.0395010395010393,
      "grad_norm": 9.023805618286133,
      "learning_rate": 4.358974358974359e-06,
      "loss": 1.1295,
      "step": 2924
    },
    {
      "epoch": 3.0405405405405403,
      "grad_norm": 8.089066505432129,
      "learning_rate": 4.356664356664357e-06,
      "loss": 0.4043,
      "step": 2925
    },
    {
      "epoch": 3.0415800415800414,
      "grad_norm": 7.827662944793701,
      "learning_rate": 4.354354354354355e-06,
      "loss": 0.1605,
      "step": 2926
    },
    {
      "epoch": 3.042619542619543,
      "grad_norm": 0.02499527856707573,
      "learning_rate": 4.352044352044352e-06,
      "loss": 0.0007,
      "step": 2927
    },
    {
      "epoch": 3.043659043659044,
      "grad_norm": 8.047962188720703,
      "learning_rate": 4.3497343497343505e-06,
      "loss": 0.6512,
      "step": 2928
    },
    {
      "epoch": 3.044698544698545,
      "grad_norm": 0.04987876117229462,
      "learning_rate": 4.347424347424348e-06,
      "loss": 0.0012,
      "step": 2929
    },
    {
      "epoch": 3.045738045738046,
      "grad_norm": 6.60944938659668,
      "learning_rate": 4.345114345114346e-06,
      "loss": 0.0833,
      "step": 2930
    },
    {
      "epoch": 3.046777546777547,
      "grad_norm": 0.04554561525583267,
      "learning_rate": 4.342804342804343e-06,
      "loss": 0.0008,
      "step": 2931
    },
    {
      "epoch": 3.047817047817048,
      "grad_norm": 5.888713836669922,
      "learning_rate": 4.340494340494341e-06,
      "loss": 0.2992,
      "step": 2932
    },
    {
      "epoch": 3.048856548856549,
      "grad_norm": 2.1652138233184814,
      "learning_rate": 4.338184338184339e-06,
      "loss": 0.0709,
      "step": 2933
    },
    {
      "epoch": 3.04989604989605,
      "grad_norm": 4.219902515411377,
      "learning_rate": 4.335874335874336e-06,
      "loss": 0.1107,
      "step": 2934
    },
    {
      "epoch": 3.050935550935551,
      "grad_norm": 0.6310142874717712,
      "learning_rate": 4.3335643335643344e-06,
      "loss": 0.0095,
      "step": 2935
    },
    {
      "epoch": 3.051975051975052,
      "grad_norm": 0.9771106839179993,
      "learning_rate": 4.331254331254332e-06,
      "loss": 0.0239,
      "step": 2936
    },
    {
      "epoch": 3.053014553014553,
      "grad_norm": 8.032059669494629,
      "learning_rate": 4.328944328944329e-06,
      "loss": 0.3691,
      "step": 2937
    },
    {
      "epoch": 3.054054054054054,
      "grad_norm": 16.941720962524414,
      "learning_rate": 4.326634326634327e-06,
      "loss": 2.0285,
      "step": 2938
    },
    {
      "epoch": 3.055093555093555,
      "grad_norm": 1.6030668020248413,
      "learning_rate": 4.324324324324325e-06,
      "loss": 0.0434,
      "step": 2939
    },
    {
      "epoch": 3.056133056133056,
      "grad_norm": 0.09329555928707123,
      "learning_rate": 4.322014322014322e-06,
      "loss": 0.0016,
      "step": 2940
    },
    {
      "epoch": 3.057172557172557,
      "grad_norm": 0.026322314515709877,
      "learning_rate": 4.31970431970432e-06,
      "loss": 0.0007,
      "step": 2941
    },
    {
      "epoch": 3.0582120582120584,
      "grad_norm": 1.3071532249450684,
      "learning_rate": 4.3173943173943175e-06,
      "loss": 0.0296,
      "step": 2942
    },
    {
      "epoch": 3.0592515592515594,
      "grad_norm": 0.10186092555522919,
      "learning_rate": 4.315084315084315e-06,
      "loss": 0.0022,
      "step": 2943
    },
    {
      "epoch": 3.0602910602910605,
      "grad_norm": 3.7780628204345703,
      "learning_rate": 4.312774312774313e-06,
      "loss": 0.1064,
      "step": 2944
    },
    {
      "epoch": 3.0613305613305615,
      "grad_norm": 2.3046798706054688,
      "learning_rate": 4.31046431046431e-06,
      "loss": 0.2597,
      "step": 2945
    },
    {
      "epoch": 3.0623700623700625,
      "grad_norm": 5.836421966552734,
      "learning_rate": 4.308154308154309e-06,
      "loss": 0.3987,
      "step": 2946
    },
    {
      "epoch": 3.0634095634095635,
      "grad_norm": 0.18187105655670166,
      "learning_rate": 4.305844305844306e-06,
      "loss": 0.0068,
      "step": 2947
    },
    {
      "epoch": 3.0644490644490645,
      "grad_norm": 0.03847048431634903,
      "learning_rate": 4.303534303534304e-06,
      "loss": 0.0011,
      "step": 2948
    },
    {
      "epoch": 3.0654885654885655,
      "grad_norm": 0.19393624365329742,
      "learning_rate": 4.3012243012243015e-06,
      "loss": 0.0052,
      "step": 2949
    },
    {
      "epoch": 3.0665280665280665,
      "grad_norm": 11.248908996582031,
      "learning_rate": 4.298914298914299e-06,
      "loss": 0.4773,
      "step": 2950
    },
    {
      "epoch": 3.0675675675675675,
      "grad_norm": 0.041529685258865356,
      "learning_rate": 4.296604296604297e-06,
      "loss": 0.0007,
      "step": 2951
    },
    {
      "epoch": 3.0686070686070686,
      "grad_norm": 17.413583755493164,
      "learning_rate": 4.294294294294294e-06,
      "loss": 0.7445,
      "step": 2952
    },
    {
      "epoch": 3.0696465696465696,
      "grad_norm": 0.8456288576126099,
      "learning_rate": 4.2919842919842925e-06,
      "loss": 0.0247,
      "step": 2953
    },
    {
      "epoch": 3.0706860706860706,
      "grad_norm": 7.395755290985107,
      "learning_rate": 4.28967428967429e-06,
      "loss": 0.3355,
      "step": 2954
    },
    {
      "epoch": 3.0717255717255716,
      "grad_norm": 6.521220684051514,
      "learning_rate": 4.287364287364288e-06,
      "loss": 0.2584,
      "step": 2955
    },
    {
      "epoch": 3.0727650727650726,
      "grad_norm": 0.22646839916706085,
      "learning_rate": 4.2850542850542854e-06,
      "loss": 0.002,
      "step": 2956
    },
    {
      "epoch": 3.0738045738045736,
      "grad_norm": 17.819766998291016,
      "learning_rate": 4.282744282744284e-06,
      "loss": 0.9187,
      "step": 2957
    },
    {
      "epoch": 3.0748440748440746,
      "grad_norm": 4.8948564529418945,
      "learning_rate": 4.280434280434281e-06,
      "loss": 0.1083,
      "step": 2958
    },
    {
      "epoch": 3.075883575883576,
      "grad_norm": 5.55280065536499,
      "learning_rate": 4.278124278124278e-06,
      "loss": 0.1584,
      "step": 2959
    },
    {
      "epoch": 3.076923076923077,
      "grad_norm": 9.615129470825195,
      "learning_rate": 4.2758142758142765e-06,
      "loss": 1.0435,
      "step": 2960
    },
    {
      "epoch": 3.077962577962578,
      "grad_norm": 5.408070087432861,
      "learning_rate": 4.273504273504274e-06,
      "loss": 0.1181,
      "step": 2961
    },
    {
      "epoch": 3.079002079002079,
      "grad_norm": 16.333675384521484,
      "learning_rate": 4.271194271194272e-06,
      "loss": 0.9409,
      "step": 2962
    },
    {
      "epoch": 3.08004158004158,
      "grad_norm": 0.8769080638885498,
      "learning_rate": 4.268884268884269e-06,
      "loss": 0.0207,
      "step": 2963
    },
    {
      "epoch": 3.081081081081081,
      "grad_norm": 4.57643985748291,
      "learning_rate": 4.266574266574267e-06,
      "loss": 0.1731,
      "step": 2964
    },
    {
      "epoch": 3.082120582120582,
      "grad_norm": 5.595919132232666,
      "learning_rate": 4.264264264264265e-06,
      "loss": 0.5219,
      "step": 2965
    },
    {
      "epoch": 3.083160083160083,
      "grad_norm": 0.01822960004210472,
      "learning_rate": 4.261954261954262e-06,
      "loss": 0.0006,
      "step": 2966
    },
    {
      "epoch": 3.084199584199584,
      "grad_norm": 4.2607102394104,
      "learning_rate": 4.25964425964426e-06,
      "loss": 0.2796,
      "step": 2967
    },
    {
      "epoch": 3.085239085239085,
      "grad_norm": 0.007498827762901783,
      "learning_rate": 4.257334257334258e-06,
      "loss": 0.0002,
      "step": 2968
    },
    {
      "epoch": 3.086278586278586,
      "grad_norm": 1.3731168508529663,
      "learning_rate": 4.255024255024255e-06,
      "loss": 0.0231,
      "step": 2969
    },
    {
      "epoch": 3.087318087318087,
      "grad_norm": 1.0861774682998657,
      "learning_rate": 4.2527142527142525e-06,
      "loss": 0.0377,
      "step": 2970
    },
    {
      "epoch": 3.0883575883575882,
      "grad_norm": 0.18928325176239014,
      "learning_rate": 4.250404250404251e-06,
      "loss": 0.005,
      "step": 2971
    },
    {
      "epoch": 3.0893970893970892,
      "grad_norm": 11.383058547973633,
      "learning_rate": 4.248094248094248e-06,
      "loss": 0.6004,
      "step": 2972
    },
    {
      "epoch": 3.0904365904365902,
      "grad_norm": 1.9454607963562012,
      "learning_rate": 4.245784245784246e-06,
      "loss": 0.0708,
      "step": 2973
    },
    {
      "epoch": 3.0914760914760917,
      "grad_norm": 7.079811096191406,
      "learning_rate": 4.2434742434742435e-06,
      "loss": 0.1478,
      "step": 2974
    },
    {
      "epoch": 3.0925155925155927,
      "grad_norm": 0.37563154101371765,
      "learning_rate": 4.241164241164242e-06,
      "loss": 0.0132,
      "step": 2975
    },
    {
      "epoch": 3.0935550935550937,
      "grad_norm": 10.523004531860352,
      "learning_rate": 4.238854238854239e-06,
      "loss": 0.8742,
      "step": 2976
    },
    {
      "epoch": 3.0945945945945947,
      "grad_norm": 2.2287039756774902,
      "learning_rate": 4.236544236544236e-06,
      "loss": 0.0512,
      "step": 2977
    },
    {
      "epoch": 3.0956340956340958,
      "grad_norm": 4.583277225494385,
      "learning_rate": 4.234234234234235e-06,
      "loss": 0.2034,
      "step": 2978
    },
    {
      "epoch": 3.0966735966735968,
      "grad_norm": 0.5153438448905945,
      "learning_rate": 4.231924231924232e-06,
      "loss": 0.0113,
      "step": 2979
    },
    {
      "epoch": 3.0977130977130978,
      "grad_norm": 12.180269241333008,
      "learning_rate": 4.22961422961423e-06,
      "loss": 0.6624,
      "step": 2980
    },
    {
      "epoch": 3.098752598752599,
      "grad_norm": 0.4038905203342438,
      "learning_rate": 4.2273042273042275e-06,
      "loss": 0.0079,
      "step": 2981
    },
    {
      "epoch": 3.0997920997921,
      "grad_norm": 0.38581734895706177,
      "learning_rate": 4.224994224994226e-06,
      "loss": 0.0049,
      "step": 2982
    },
    {
      "epoch": 3.100831600831601,
      "grad_norm": 6.3126420974731445,
      "learning_rate": 4.222684222684223e-06,
      "loss": 0.3044,
      "step": 2983
    },
    {
      "epoch": 3.101871101871102,
      "grad_norm": 9.254770278930664,
      "learning_rate": 4.220374220374221e-06,
      "loss": 0.6881,
      "step": 2984
    },
    {
      "epoch": 3.102910602910603,
      "grad_norm": 0.7424444556236267,
      "learning_rate": 4.2180642180642186e-06,
      "loss": 0.0077,
      "step": 2985
    },
    {
      "epoch": 3.103950103950104,
      "grad_norm": 5.989633083343506,
      "learning_rate": 4.215754215754216e-06,
      "loss": 0.1914,
      "step": 2986
    },
    {
      "epoch": 3.104989604989605,
      "grad_norm": 4.006546974182129,
      "learning_rate": 4.213444213444214e-06,
      "loss": 0.0936,
      "step": 2987
    },
    {
      "epoch": 3.106029106029106,
      "grad_norm": 0.39120224118232727,
      "learning_rate": 4.2111342111342114e-06,
      "loss": 0.0091,
      "step": 2988
    },
    {
      "epoch": 3.107068607068607,
      "grad_norm": 7.164828300476074,
      "learning_rate": 4.20882420882421e-06,
      "loss": 0.3958,
      "step": 2989
    },
    {
      "epoch": 3.108108108108108,
      "grad_norm": 1.572396159172058,
      "learning_rate": 4.206514206514207e-06,
      "loss": 0.0273,
      "step": 2990
    },
    {
      "epoch": 3.1091476091476093,
      "grad_norm": 0.8376824259757996,
      "learning_rate": 4.204204204204204e-06,
      "loss": 0.2671,
      "step": 2991
    },
    {
      "epoch": 3.1101871101871104,
      "grad_norm": 8.45645523071289,
      "learning_rate": 4.2018942018942025e-06,
      "loss": 0.5827,
      "step": 2992
    },
    {
      "epoch": 3.1112266112266114,
      "grad_norm": 6.415794372558594,
      "learning_rate": 4.1995841995842e-06,
      "loss": 0.2154,
      "step": 2993
    },
    {
      "epoch": 3.1122661122661124,
      "grad_norm": 2.558643341064453,
      "learning_rate": 4.197274197274197e-06,
      "loss": 0.0716,
      "step": 2994
    },
    {
      "epoch": 3.1133056133056134,
      "grad_norm": 0.21552839875221252,
      "learning_rate": 4.194964194964195e-06,
      "loss": 0.0052,
      "step": 2995
    },
    {
      "epoch": 3.1143451143451144,
      "grad_norm": 8.271615028381348,
      "learning_rate": 4.192654192654193e-06,
      "loss": 0.2212,
      "step": 2996
    },
    {
      "epoch": 3.1153846153846154,
      "grad_norm": 5.949480056762695,
      "learning_rate": 4.19034419034419e-06,
      "loss": 0.3214,
      "step": 2997
    },
    {
      "epoch": 3.1164241164241164,
      "grad_norm": 4.731964588165283,
      "learning_rate": 4.188034188034188e-06,
      "loss": 0.1788,
      "step": 2998
    },
    {
      "epoch": 3.1174636174636174,
      "grad_norm": 11.179065704345703,
      "learning_rate": 4.185724185724186e-06,
      "loss": 0.4589,
      "step": 2999
    },
    {
      "epoch": 3.1185031185031185,
      "grad_norm": 6.513021469116211,
      "learning_rate": 4.183414183414184e-06,
      "loss": 0.2555,
      "step": 3000
    },
    {
      "epoch": 3.1195426195426195,
      "grad_norm": 4.68575382232666,
      "learning_rate": 4.181104181104181e-06,
      "loss": 0.1547,
      "step": 3001
    },
    {
      "epoch": 3.1205821205821205,
      "grad_norm": 0.18110813200473785,
      "learning_rate": 4.178794178794179e-06,
      "loss": 0.0055,
      "step": 3002
    },
    {
      "epoch": 3.1216216216216215,
      "grad_norm": 8.774043083190918,
      "learning_rate": 4.176484176484177e-06,
      "loss": 0.3781,
      "step": 3003
    },
    {
      "epoch": 3.1226611226611225,
      "grad_norm": 0.7165976166725159,
      "learning_rate": 4.174174174174174e-06,
      "loss": 0.0105,
      "step": 3004
    },
    {
      "epoch": 3.1237006237006235,
      "grad_norm": 9.499658584594727,
      "learning_rate": 4.171864171864172e-06,
      "loss": 0.7248,
      "step": 3005
    },
    {
      "epoch": 3.124740124740125,
      "grad_norm": 12.022478103637695,
      "learning_rate": 4.1695541695541696e-06,
      "loss": 0.8418,
      "step": 3006
    },
    {
      "epoch": 3.125779625779626,
      "grad_norm": 0.8170308470726013,
      "learning_rate": 4.167244167244168e-06,
      "loss": 0.0184,
      "step": 3007
    },
    {
      "epoch": 3.126819126819127,
      "grad_norm": 10.470986366271973,
      "learning_rate": 4.164934164934165e-06,
      "loss": 0.7221,
      "step": 3008
    },
    {
      "epoch": 3.127858627858628,
      "grad_norm": 0.19149138033390045,
      "learning_rate": 4.162624162624163e-06,
      "loss": 0.0038,
      "step": 3009
    },
    {
      "epoch": 3.128898128898129,
      "grad_norm": 7.027655124664307,
      "learning_rate": 4.160314160314161e-06,
      "loss": 0.1312,
      "step": 3010
    },
    {
      "epoch": 3.12993762993763,
      "grad_norm": 1.6696555614471436,
      "learning_rate": 4.158004158004159e-06,
      "loss": 0.027,
      "step": 3011
    },
    {
      "epoch": 3.130977130977131,
      "grad_norm": 0.03533538803458214,
      "learning_rate": 4.155694155694156e-06,
      "loss": 0.0005,
      "step": 3012
    },
    {
      "epoch": 3.132016632016632,
      "grad_norm": 1.7083377838134766,
      "learning_rate": 4.1533841533841535e-06,
      "loss": 0.045,
      "step": 3013
    },
    {
      "epoch": 3.133056133056133,
      "grad_norm": 0.1288733333349228,
      "learning_rate": 4.151074151074152e-06,
      "loss": 0.0049,
      "step": 3014
    },
    {
      "epoch": 3.134095634095634,
      "grad_norm": 6.5716233253479,
      "learning_rate": 4.148764148764149e-06,
      "loss": 0.3688,
      "step": 3015
    },
    {
      "epoch": 3.135135135135135,
      "grad_norm": 10.444952011108398,
      "learning_rate": 4.146454146454147e-06,
      "loss": 1.0296,
      "step": 3016
    },
    {
      "epoch": 3.136174636174636,
      "grad_norm": 0.9561419486999512,
      "learning_rate": 4.1441441441441446e-06,
      "loss": 0.0408,
      "step": 3017
    },
    {
      "epoch": 3.137214137214137,
      "grad_norm": 5.194184303283691,
      "learning_rate": 4.141834141834143e-06,
      "loss": 0.1648,
      "step": 3018
    },
    {
      "epoch": 3.138253638253638,
      "grad_norm": 3.6494336128234863,
      "learning_rate": 4.13952413952414e-06,
      "loss": 0.1334,
      "step": 3019
    },
    {
      "epoch": 3.139293139293139,
      "grad_norm": 3.497851848602295,
      "learning_rate": 4.1372141372141374e-06,
      "loss": 0.0748,
      "step": 3020
    },
    {
      "epoch": 3.14033264033264,
      "grad_norm": 9.248802185058594,
      "learning_rate": 4.134904134904136e-06,
      "loss": 0.342,
      "step": 3021
    },
    {
      "epoch": 3.141372141372141,
      "grad_norm": 6.574548721313477,
      "learning_rate": 4.132594132594133e-06,
      "loss": 0.24,
      "step": 3022
    },
    {
      "epoch": 3.1424116424116426,
      "grad_norm": 6.959147930145264,
      "learning_rate": 4.13028413028413e-06,
      "loss": 0.2583,
      "step": 3023
    },
    {
      "epoch": 3.1434511434511436,
      "grad_norm": 9.682708740234375,
      "learning_rate": 4.1279741279741285e-06,
      "loss": 0.4094,
      "step": 3024
    },
    {
      "epoch": 3.1444906444906446,
      "grad_norm": 0.3175424635410309,
      "learning_rate": 4.125664125664126e-06,
      "loss": 0.011,
      "step": 3025
    },
    {
      "epoch": 3.1455301455301456,
      "grad_norm": 3.7239432334899902,
      "learning_rate": 4.123354123354123e-06,
      "loss": 0.0823,
      "step": 3026
    },
    {
      "epoch": 3.1465696465696467,
      "grad_norm": 0.09776545315980911,
      "learning_rate": 4.121044121044121e-06,
      "loss": 0.0023,
      "step": 3027
    },
    {
      "epoch": 3.1476091476091477,
      "grad_norm": 8.206098556518555,
      "learning_rate": 4.118734118734119e-06,
      "loss": 0.2836,
      "step": 3028
    },
    {
      "epoch": 3.1486486486486487,
      "grad_norm": 5.508115291595459,
      "learning_rate": 4.116424116424117e-06,
      "loss": 0.1995,
      "step": 3029
    },
    {
      "epoch": 3.1496881496881497,
      "grad_norm": 6.341372966766357,
      "learning_rate": 4.114114114114114e-06,
      "loss": 0.1132,
      "step": 3030
    },
    {
      "epoch": 3.1507276507276507,
      "grad_norm": 0.16759119927883148,
      "learning_rate": 4.111804111804112e-06,
      "loss": 0.0026,
      "step": 3031
    },
    {
      "epoch": 3.1517671517671517,
      "grad_norm": 0.1679067462682724,
      "learning_rate": 4.10949410949411e-06,
      "loss": 0.0039,
      "step": 3032
    },
    {
      "epoch": 3.1528066528066527,
      "grad_norm": 0.1461794376373291,
      "learning_rate": 4.107184107184107e-06,
      "loss": 0.0026,
      "step": 3033
    },
    {
      "epoch": 3.1538461538461537,
      "grad_norm": 8.840606689453125,
      "learning_rate": 4.104874104874105e-06,
      "loss": 0.3927,
      "step": 3034
    },
    {
      "epoch": 3.1548856548856548,
      "grad_norm": 9.067317008972168,
      "learning_rate": 4.102564102564103e-06,
      "loss": 0.6051,
      "step": 3035
    },
    {
      "epoch": 3.1559251559251558,
      "grad_norm": 0.08562859892845154,
      "learning_rate": 4.100254100254101e-06,
      "loss": 0.002,
      "step": 3036
    },
    {
      "epoch": 3.156964656964657,
      "grad_norm": 0.3136712908744812,
      "learning_rate": 4.097944097944098e-06,
      "loss": 0.0064,
      "step": 3037
    },
    {
      "epoch": 3.1580041580041582,
      "grad_norm": 5.612872123718262,
      "learning_rate": 4.095634095634096e-06,
      "loss": 0.2334,
      "step": 3038
    },
    {
      "epoch": 3.1590436590436592,
      "grad_norm": 1.436871886253357,
      "learning_rate": 4.093324093324094e-06,
      "loss": 0.0241,
      "step": 3039
    },
    {
      "epoch": 3.1600831600831603,
      "grad_norm": 5.758492469787598,
      "learning_rate": 4.091014091014091e-06,
      "loss": 0.1174,
      "step": 3040
    },
    {
      "epoch": 3.1611226611226613,
      "grad_norm": 1.3731693029403687,
      "learning_rate": 4.088704088704089e-06,
      "loss": 0.0247,
      "step": 3041
    },
    {
      "epoch": 3.1621621621621623,
      "grad_norm": 4.757385730743408,
      "learning_rate": 4.086394086394087e-06,
      "loss": 0.1087,
      "step": 3042
    },
    {
      "epoch": 3.1632016632016633,
      "grad_norm": 0.881007730960846,
      "learning_rate": 4.084084084084085e-06,
      "loss": 0.02,
      "step": 3043
    },
    {
      "epoch": 3.1642411642411643,
      "grad_norm": 2.4619946479797363,
      "learning_rate": 4.081774081774082e-06,
      "loss": 0.1007,
      "step": 3044
    },
    {
      "epoch": 3.1652806652806653,
      "grad_norm": 0.38008901476860046,
      "learning_rate": 4.07946407946408e-06,
      "loss": 0.0046,
      "step": 3045
    },
    {
      "epoch": 3.1663201663201663,
      "grad_norm": 0.6488372683525085,
      "learning_rate": 4.077154077154078e-06,
      "loss": 0.0187,
      "step": 3046
    },
    {
      "epoch": 3.1673596673596673,
      "grad_norm": 3.645965576171875,
      "learning_rate": 4.074844074844075e-06,
      "loss": 0.1018,
      "step": 3047
    },
    {
      "epoch": 3.1683991683991684,
      "grad_norm": 0.9687808752059937,
      "learning_rate": 4.072534072534073e-06,
      "loss": 0.0168,
      "step": 3048
    },
    {
      "epoch": 3.1694386694386694,
      "grad_norm": 0.5096043348312378,
      "learning_rate": 4.0702240702240706e-06,
      "loss": 0.0081,
      "step": 3049
    },
    {
      "epoch": 3.1704781704781704,
      "grad_norm": 0.23202700912952423,
      "learning_rate": 4.067914067914068e-06,
      "loss": 0.0056,
      "step": 3050
    },
    {
      "epoch": 3.1715176715176714,
      "grad_norm": 3.7122135162353516,
      "learning_rate": 4.065604065604066e-06,
      "loss": 0.0737,
      "step": 3051
    },
    {
      "epoch": 3.1725571725571724,
      "grad_norm": 0.19520847499370575,
      "learning_rate": 4.0632940632940635e-06,
      "loss": 0.006,
      "step": 3052
    },
    {
      "epoch": 3.1735966735966734,
      "grad_norm": 2.914714813232422,
      "learning_rate": 4.060984060984061e-06,
      "loss": 0.0777,
      "step": 3053
    },
    {
      "epoch": 3.1746361746361744,
      "grad_norm": 2.344712018966675,
      "learning_rate": 4.058674058674059e-06,
      "loss": 0.0649,
      "step": 3054
    },
    {
      "epoch": 3.175675675675676,
      "grad_norm": 7.863626003265381,
      "learning_rate": 4.056364056364056e-06,
      "loss": 0.1842,
      "step": 3055
    },
    {
      "epoch": 3.176715176715177,
      "grad_norm": 8.394662857055664,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 0.5036,
      "step": 3056
    },
    {
      "epoch": 3.177754677754678,
      "grad_norm": 4.316769599914551,
      "learning_rate": 4.051744051744052e-06,
      "loss": 0.0995,
      "step": 3057
    },
    {
      "epoch": 3.178794178794179,
      "grad_norm": 4.495481967926025,
      "learning_rate": 4.049434049434049e-06,
      "loss": 0.1215,
      "step": 3058
    },
    {
      "epoch": 3.17983367983368,
      "grad_norm": 9.788843154907227,
      "learning_rate": 4.047124047124047e-06,
      "loss": 1.0258,
      "step": 3059
    },
    {
      "epoch": 3.180873180873181,
      "grad_norm": 9.788480758666992,
      "learning_rate": 4.044814044814045e-06,
      "loss": 0.7014,
      "step": 3060
    },
    {
      "epoch": 3.181912681912682,
      "grad_norm": 14.928915977478027,
      "learning_rate": 4.042504042504043e-06,
      "loss": 1.2652,
      "step": 3061
    },
    {
      "epoch": 3.182952182952183,
      "grad_norm": 13.048415184020996,
      "learning_rate": 4.04019404019404e-06,
      "loss": 0.3149,
      "step": 3062
    },
    {
      "epoch": 3.183991683991684,
      "grad_norm": 3.889265537261963,
      "learning_rate": 4.0378840378840385e-06,
      "loss": 0.0809,
      "step": 3063
    },
    {
      "epoch": 3.185031185031185,
      "grad_norm": 0.15134768187999725,
      "learning_rate": 4.035574035574036e-06,
      "loss": 0.0026,
      "step": 3064
    },
    {
      "epoch": 3.186070686070686,
      "grad_norm": 5.497805118560791,
      "learning_rate": 4.033264033264034e-06,
      "loss": 0.2563,
      "step": 3065
    },
    {
      "epoch": 3.187110187110187,
      "grad_norm": 6.498611927032471,
      "learning_rate": 4.030954030954031e-06,
      "loss": 0.3335,
      "step": 3066
    },
    {
      "epoch": 3.188149688149688,
      "grad_norm": 12.062654495239258,
      "learning_rate": 4.028644028644029e-06,
      "loss": 0.6318,
      "step": 3067
    },
    {
      "epoch": 3.189189189189189,
      "grad_norm": 3.560054302215576,
      "learning_rate": 4.026334026334027e-06,
      "loss": 0.0463,
      "step": 3068
    },
    {
      "epoch": 3.19022869022869,
      "grad_norm": 8.133339881896973,
      "learning_rate": 4.024024024024024e-06,
      "loss": 0.5378,
      "step": 3069
    },
    {
      "epoch": 3.1912681912681915,
      "grad_norm": 6.538513660430908,
      "learning_rate": 4.021714021714022e-06,
      "loss": 0.4576,
      "step": 3070
    },
    {
      "epoch": 3.1923076923076925,
      "grad_norm": 1.1551824808120728,
      "learning_rate": 4.01940401940402e-06,
      "loss": 0.0318,
      "step": 3071
    },
    {
      "epoch": 3.1933471933471935,
      "grad_norm": 1.8024814128875732,
      "learning_rate": 4.017094017094018e-06,
      "loss": 0.056,
      "step": 3072
    },
    {
      "epoch": 3.1943866943866945,
      "grad_norm": 12.158453941345215,
      "learning_rate": 4.014784014784015e-06,
      "loss": 0.1438,
      "step": 3073
    },
    {
      "epoch": 3.1954261954261955,
      "grad_norm": 0.8606366515159607,
      "learning_rate": 4.012474012474013e-06,
      "loss": 0.0071,
      "step": 3074
    },
    {
      "epoch": 3.1964656964656966,
      "grad_norm": 5.564821720123291,
      "learning_rate": 4.010164010164011e-06,
      "loss": 0.1433,
      "step": 3075
    },
    {
      "epoch": 3.1975051975051976,
      "grad_norm": 4.401222229003906,
      "learning_rate": 4.007854007854008e-06,
      "loss": 0.1796,
      "step": 3076
    },
    {
      "epoch": 3.1985446985446986,
      "grad_norm": 0.0049988687969744205,
      "learning_rate": 4.0055440055440055e-06,
      "loss": 0.0001,
      "step": 3077
    },
    {
      "epoch": 3.1995841995841996,
      "grad_norm": 3.3277342319488525,
      "learning_rate": 4.003234003234004e-06,
      "loss": 0.0776,
      "step": 3078
    },
    {
      "epoch": 3.2006237006237006,
      "grad_norm": 5.566887855529785,
      "learning_rate": 4.000924000924001e-06,
      "loss": 0.1188,
      "step": 3079
    },
    {
      "epoch": 3.2016632016632016,
      "grad_norm": 7.9430928230285645,
      "learning_rate": 3.998613998613998e-06,
      "loss": 0.2335,
      "step": 3080
    },
    {
      "epoch": 3.2027027027027026,
      "grad_norm": 7.287680625915527,
      "learning_rate": 3.996303996303997e-06,
      "loss": 0.2504,
      "step": 3081
    },
    {
      "epoch": 3.2037422037422036,
      "grad_norm": 2.037623405456543,
      "learning_rate": 3.993993993993994e-06,
      "loss": 0.0641,
      "step": 3082
    },
    {
      "epoch": 3.2047817047817047,
      "grad_norm": 8.476522445678711,
      "learning_rate": 3.991683991683992e-06,
      "loss": 0.454,
      "step": 3083
    },
    {
      "epoch": 3.2058212058212057,
      "grad_norm": 11.712259292602539,
      "learning_rate": 3.9893739893739895e-06,
      "loss": 0.6441,
      "step": 3084
    },
    {
      "epoch": 3.2068607068607067,
      "grad_norm": 0.12107273191213608,
      "learning_rate": 3.987063987063987e-06,
      "loss": 0.0024,
      "step": 3085
    },
    {
      "epoch": 3.2079002079002077,
      "grad_norm": 0.0172722265124321,
      "learning_rate": 3.984753984753985e-06,
      "loss": 0.0003,
      "step": 3086
    },
    {
      "epoch": 3.208939708939709,
      "grad_norm": 7.512850284576416,
      "learning_rate": 3.982443982443982e-06,
      "loss": 0.2886,
      "step": 3087
    },
    {
      "epoch": 3.20997920997921,
      "grad_norm": 10.699891090393066,
      "learning_rate": 3.9801339801339805e-06,
      "loss": 1.306,
      "step": 3088
    },
    {
      "epoch": 3.211018711018711,
      "grad_norm": 3.024954080581665,
      "learning_rate": 3.977823977823978e-06,
      "loss": 0.2697,
      "step": 3089
    },
    {
      "epoch": 3.212058212058212,
      "grad_norm": 8.047675132751465,
      "learning_rate": 3.975513975513976e-06,
      "loss": 0.2475,
      "step": 3090
    },
    {
      "epoch": 3.213097713097713,
      "grad_norm": 0.10544872283935547,
      "learning_rate": 3.973203973203973e-06,
      "loss": 0.0018,
      "step": 3091
    },
    {
      "epoch": 3.214137214137214,
      "grad_norm": 0.11675427854061127,
      "learning_rate": 3.970893970893972e-06,
      "loss": 0.0015,
      "step": 3092
    },
    {
      "epoch": 3.215176715176715,
      "grad_norm": 13.126848220825195,
      "learning_rate": 3.968583968583969e-06,
      "loss": 0.4408,
      "step": 3093
    },
    {
      "epoch": 3.2162162162162162,
      "grad_norm": 12.223016738891602,
      "learning_rate": 3.966273966273966e-06,
      "loss": 0.0774,
      "step": 3094
    },
    {
      "epoch": 3.2172557172557172,
      "grad_norm": 0.5851736664772034,
      "learning_rate": 3.9639639639639645e-06,
      "loss": 0.0133,
      "step": 3095
    },
    {
      "epoch": 3.2182952182952183,
      "grad_norm": 4.598794937133789,
      "learning_rate": 3.961653961653962e-06,
      "loss": 0.0648,
      "step": 3096
    },
    {
      "epoch": 3.2193347193347193,
      "grad_norm": 4.362109184265137,
      "learning_rate": 3.95934395934396e-06,
      "loss": 0.2351,
      "step": 3097
    },
    {
      "epoch": 3.2203742203742203,
      "grad_norm": 15.790438652038574,
      "learning_rate": 3.957033957033957e-06,
      "loss": 0.2569,
      "step": 3098
    },
    {
      "epoch": 3.2214137214137213,
      "grad_norm": 9.33132266998291,
      "learning_rate": 3.9547239547239555e-06,
      "loss": 0.4077,
      "step": 3099
    },
    {
      "epoch": 3.2224532224532223,
      "grad_norm": 1.6285666227340698,
      "learning_rate": 3.952413952413953e-06,
      "loss": 0.0337,
      "step": 3100
    },
    {
      "epoch": 3.2234927234927233,
      "grad_norm": 1.7507997751235962,
      "learning_rate": 3.95010395010395e-06,
      "loss": 0.0333,
      "step": 3101
    },
    {
      "epoch": 3.2245322245322248,
      "grad_norm": 1.823251724243164,
      "learning_rate": 3.9477939477939484e-06,
      "loss": 0.0389,
      "step": 3102
    },
    {
      "epoch": 3.225571725571726,
      "grad_norm": 4.533589839935303,
      "learning_rate": 3.945483945483946e-06,
      "loss": 0.2476,
      "step": 3103
    },
    {
      "epoch": 3.226611226611227,
      "grad_norm": 0.8647873401641846,
      "learning_rate": 3.943173943173943e-06,
      "loss": 0.0198,
      "step": 3104
    },
    {
      "epoch": 3.227650727650728,
      "grad_norm": 9.659738540649414,
      "learning_rate": 3.940863940863941e-06,
      "loss": 0.5419,
      "step": 3105
    },
    {
      "epoch": 3.228690228690229,
      "grad_norm": 0.054894059896469116,
      "learning_rate": 3.938553938553939e-06,
      "loss": 0.0015,
      "step": 3106
    },
    {
      "epoch": 3.22972972972973,
      "grad_norm": 1.3090507984161377,
      "learning_rate": 3.936243936243936e-06,
      "loss": 0.0199,
      "step": 3107
    },
    {
      "epoch": 3.230769230769231,
      "grad_norm": 5.187557697296143,
      "learning_rate": 3.933933933933934e-06,
      "loss": 0.1225,
      "step": 3108
    },
    {
      "epoch": 3.231808731808732,
      "grad_norm": 0.2481028288602829,
      "learning_rate": 3.9316239316239315e-06,
      "loss": 0.0065,
      "step": 3109
    },
    {
      "epoch": 3.232848232848233,
      "grad_norm": 10.085977554321289,
      "learning_rate": 3.92931392931393e-06,
      "loss": 0.4734,
      "step": 3110
    },
    {
      "epoch": 3.233887733887734,
      "grad_norm": 0.2195979505777359,
      "learning_rate": 3.927003927003927e-06,
      "loss": 0.0066,
      "step": 3111
    },
    {
      "epoch": 3.234927234927235,
      "grad_norm": 0.9394448399543762,
      "learning_rate": 3.924693924693924e-06,
      "loss": 0.032,
      "step": 3112
    },
    {
      "epoch": 3.235966735966736,
      "grad_norm": 0.003855444025248289,
      "learning_rate": 3.922383922383923e-06,
      "loss": 0.0001,
      "step": 3113
    },
    {
      "epoch": 3.237006237006237,
      "grad_norm": 2.7753994464874268,
      "learning_rate": 3.92007392007392e-06,
      "loss": 0.0534,
      "step": 3114
    },
    {
      "epoch": 3.238045738045738,
      "grad_norm": 2.2066776752471924,
      "learning_rate": 3.917763917763918e-06,
      "loss": 0.06,
      "step": 3115
    },
    {
      "epoch": 3.239085239085239,
      "grad_norm": 3.437779426574707,
      "learning_rate": 3.9154539154539155e-06,
      "loss": 0.0524,
      "step": 3116
    },
    {
      "epoch": 3.24012474012474,
      "grad_norm": 0.572167694568634,
      "learning_rate": 3.913143913143914e-06,
      "loss": 0.013,
      "step": 3117
    },
    {
      "epoch": 3.241164241164241,
      "grad_norm": 8.824238777160645,
      "learning_rate": 3.910833910833911e-06,
      "loss": 0.4768,
      "step": 3118
    },
    {
      "epoch": 3.2422037422037424,
      "grad_norm": 7.6617584228515625,
      "learning_rate": 3.908523908523909e-06,
      "loss": 0.4057,
      "step": 3119
    },
    {
      "epoch": 3.2432432432432434,
      "grad_norm": 0.5249945521354675,
      "learning_rate": 3.9062139062139065e-06,
      "loss": 0.009,
      "step": 3120
    },
    {
      "epoch": 3.2442827442827444,
      "grad_norm": 1.2607581615447998,
      "learning_rate": 3.903903903903904e-06,
      "loss": 0.0261,
      "step": 3121
    },
    {
      "epoch": 3.2453222453222454,
      "grad_norm": 2.4366354942321777,
      "learning_rate": 3.901593901593902e-06,
      "loss": 0.0777,
      "step": 3122
    },
    {
      "epoch": 3.2463617463617465,
      "grad_norm": 10.035798072814941,
      "learning_rate": 3.899283899283899e-06,
      "loss": 0.4241,
      "step": 3123
    },
    {
      "epoch": 3.2474012474012475,
      "grad_norm": 6.8840413093566895,
      "learning_rate": 3.896973896973898e-06,
      "loss": 0.2189,
      "step": 3124
    },
    {
      "epoch": 3.2484407484407485,
      "grad_norm": 2.9253406524658203,
      "learning_rate": 3.894663894663895e-06,
      "loss": 0.0927,
      "step": 3125
    },
    {
      "epoch": 3.2494802494802495,
      "grad_norm": 3.3498616218566895,
      "learning_rate": 3.892353892353893e-06,
      "loss": 0.0548,
      "step": 3126
    },
    {
      "epoch": 3.2505197505197505,
      "grad_norm": 0.8399617671966553,
      "learning_rate": 3.8900438900438905e-06,
      "loss": 0.0177,
      "step": 3127
    },
    {
      "epoch": 3.2515592515592515,
      "grad_norm": 1.4739288091659546,
      "learning_rate": 3.887733887733889e-06,
      "loss": 0.0303,
      "step": 3128
    },
    {
      "epoch": 3.2525987525987525,
      "grad_norm": 7.085636615753174,
      "learning_rate": 3.885423885423886e-06,
      "loss": 0.4284,
      "step": 3129
    },
    {
      "epoch": 3.2536382536382535,
      "grad_norm": 4.9388933181762695,
      "learning_rate": 3.883113883113883e-06,
      "loss": 0.1027,
      "step": 3130
    },
    {
      "epoch": 3.2546777546777546,
      "grad_norm": 8.358454704284668,
      "learning_rate": 3.8808038808038816e-06,
      "loss": 0.8982,
      "step": 3131
    },
    {
      "epoch": 3.2557172557172556,
      "grad_norm": 2.0921363830566406,
      "learning_rate": 3.878493878493879e-06,
      "loss": 0.0371,
      "step": 3132
    },
    {
      "epoch": 3.2567567567567566,
      "grad_norm": 1.1191800832748413,
      "learning_rate": 3.876183876183876e-06,
      "loss": 0.2502,
      "step": 3133
    },
    {
      "epoch": 3.257796257796258,
      "grad_norm": 1.5055001974105835,
      "learning_rate": 3.8738738738738744e-06,
      "loss": 0.0246,
      "step": 3134
    },
    {
      "epoch": 3.258835758835759,
      "grad_norm": 20.0693302154541,
      "learning_rate": 3.871563871563872e-06,
      "loss": 1.0522,
      "step": 3135
    },
    {
      "epoch": 3.25987525987526,
      "grad_norm": 0.03470122814178467,
      "learning_rate": 3.869253869253869e-06,
      "loss": 0.0005,
      "step": 3136
    },
    {
      "epoch": 3.260914760914761,
      "grad_norm": 7.206092834472656,
      "learning_rate": 3.866943866943867e-06,
      "loss": 1.1063,
      "step": 3137
    },
    {
      "epoch": 3.261954261954262,
      "grad_norm": 1.0938373804092407,
      "learning_rate": 3.864633864633865e-06,
      "loss": 0.024,
      "step": 3138
    },
    {
      "epoch": 3.262993762993763,
      "grad_norm": 4.277055263519287,
      "learning_rate": 3.862323862323863e-06,
      "loss": 0.1911,
      "step": 3139
    },
    {
      "epoch": 3.264033264033264,
      "grad_norm": 0.10324853658676147,
      "learning_rate": 3.86001386001386e-06,
      "loss": 0.0024,
      "step": 3140
    },
    {
      "epoch": 3.265072765072765,
      "grad_norm": 3.6633353233337402,
      "learning_rate": 3.8577038577038575e-06,
      "loss": 0.0671,
      "step": 3141
    },
    {
      "epoch": 3.266112266112266,
      "grad_norm": 2.2221670150756836,
      "learning_rate": 3.855393855393856e-06,
      "loss": 0.0271,
      "step": 3142
    },
    {
      "epoch": 3.267151767151767,
      "grad_norm": 0.14709371328353882,
      "learning_rate": 3.853083853083853e-06,
      "loss": 0.0056,
      "step": 3143
    },
    {
      "epoch": 3.268191268191268,
      "grad_norm": 4.218796730041504,
      "learning_rate": 3.850773850773851e-06,
      "loss": 0.3053,
      "step": 3144
    },
    {
      "epoch": 3.269230769230769,
      "grad_norm": 3.633521318435669,
      "learning_rate": 3.848463848463849e-06,
      "loss": 0.0535,
      "step": 3145
    },
    {
      "epoch": 3.27027027027027,
      "grad_norm": 2.324286937713623,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.0562,
      "step": 3146
    },
    {
      "epoch": 3.271309771309771,
      "grad_norm": 4.2747111320495605,
      "learning_rate": 3.843843843843844e-06,
      "loss": 0.1062,
      "step": 3147
    },
    {
      "epoch": 3.272349272349272,
      "grad_norm": 0.26570233702659607,
      "learning_rate": 3.841533841533842e-06,
      "loss": 0.0053,
      "step": 3148
    },
    {
      "epoch": 3.273388773388773,
      "grad_norm": 0.3976437449455261,
      "learning_rate": 3.83922383922384e-06,
      "loss": 0.0061,
      "step": 3149
    },
    {
      "epoch": 3.274428274428274,
      "grad_norm": 0.36365818977355957,
      "learning_rate": 3.836913836913837e-06,
      "loss": 0.0093,
      "step": 3150
    },
    {
      "epoch": 3.2754677754677752,
      "grad_norm": 4.533632755279541,
      "learning_rate": 3.834603834603835e-06,
      "loss": 0.3268,
      "step": 3151
    },
    {
      "epoch": 3.2765072765072767,
      "grad_norm": 1.8680239915847778,
      "learning_rate": 3.8322938322938326e-06,
      "loss": 0.0712,
      "step": 3152
    },
    {
      "epoch": 3.2775467775467777,
      "grad_norm": 5.484252452850342,
      "learning_rate": 3.829983829983831e-06,
      "loss": 0.1009,
      "step": 3153
    },
    {
      "epoch": 3.2785862785862787,
      "grad_norm": 9.467362403869629,
      "learning_rate": 3.827673827673828e-06,
      "loss": 0.4734,
      "step": 3154
    },
    {
      "epoch": 3.2796257796257797,
      "grad_norm": 0.005934694781899452,
      "learning_rate": 3.825363825363826e-06,
      "loss": 0.0001,
      "step": 3155
    },
    {
      "epoch": 3.2806652806652807,
      "grad_norm": 9.528066635131836,
      "learning_rate": 3.823053823053824e-06,
      "loss": 1.0359,
      "step": 3156
    },
    {
      "epoch": 3.2817047817047817,
      "grad_norm": 0.7023597359657288,
      "learning_rate": 3.820743820743821e-06,
      "loss": 0.018,
      "step": 3157
    },
    {
      "epoch": 3.2827442827442828,
      "grad_norm": 0.018386509269475937,
      "learning_rate": 3.818433818433819e-06,
      "loss": 0.0004,
      "step": 3158
    },
    {
      "epoch": 3.2837837837837838,
      "grad_norm": 1.8641653060913086,
      "learning_rate": 3.8161238161238165e-06,
      "loss": 0.038,
      "step": 3159
    },
    {
      "epoch": 3.284823284823285,
      "grad_norm": 7.680948257446289,
      "learning_rate": 3.8138138138138143e-06,
      "loss": 0.1728,
      "step": 3160
    },
    {
      "epoch": 3.285862785862786,
      "grad_norm": 5.503201007843018,
      "learning_rate": 3.8115038115038116e-06,
      "loss": 0.0966,
      "step": 3161
    },
    {
      "epoch": 3.286902286902287,
      "grad_norm": 17.328290939331055,
      "learning_rate": 3.80919380919381e-06,
      "loss": 2.5301,
      "step": 3162
    },
    {
      "epoch": 3.287941787941788,
      "grad_norm": 0.25888022780418396,
      "learning_rate": 3.806883806883807e-06,
      "loss": 0.0081,
      "step": 3163
    },
    {
      "epoch": 3.288981288981289,
      "grad_norm": 5.120408535003662,
      "learning_rate": 3.804573804573805e-06,
      "loss": 0.2536,
      "step": 3164
    },
    {
      "epoch": 3.29002079002079,
      "grad_norm": 9.060827255249023,
      "learning_rate": 3.8022638022638027e-06,
      "loss": 0.8146,
      "step": 3165
    },
    {
      "epoch": 3.2910602910602913,
      "grad_norm": 1.1167017221450806,
      "learning_rate": 3.7999537999538004e-06,
      "loss": 0.0379,
      "step": 3166
    },
    {
      "epoch": 3.2920997920997923,
      "grad_norm": 2.823840379714966,
      "learning_rate": 3.7976437976437978e-06,
      "loss": 0.0544,
      "step": 3167
    },
    {
      "epoch": 3.2931392931392933,
      "grad_norm": 4.674472808837891,
      "learning_rate": 3.7953337953337956e-06,
      "loss": 0.2274,
      "step": 3168
    },
    {
      "epoch": 3.2941787941787943,
      "grad_norm": 8.276628494262695,
      "learning_rate": 3.7930237930237933e-06,
      "loss": 0.4559,
      "step": 3169
    },
    {
      "epoch": 3.2952182952182953,
      "grad_norm": 3.9904773235321045,
      "learning_rate": 3.7907137907137907e-06,
      "loss": 0.1314,
      "step": 3170
    },
    {
      "epoch": 3.2962577962577964,
      "grad_norm": 2.9665725231170654,
      "learning_rate": 3.788403788403789e-06,
      "loss": 0.0674,
      "step": 3171
    },
    {
      "epoch": 3.2972972972972974,
      "grad_norm": 0.3282411992549896,
      "learning_rate": 3.786093786093786e-06,
      "loss": 0.0073,
      "step": 3172
    },
    {
      "epoch": 3.2983367983367984,
      "grad_norm": 3.566387414932251,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 0.0305,
      "step": 3173
    },
    {
      "epoch": 3.2993762993762994,
      "grad_norm": 0.08132052421569824,
      "learning_rate": 3.7814737814737817e-06,
      "loss": 0.0018,
      "step": 3174
    },
    {
      "epoch": 3.3004158004158004,
      "grad_norm": 0.3602537214756012,
      "learning_rate": 3.77916377916378e-06,
      "loss": 0.0076,
      "step": 3175
    },
    {
      "epoch": 3.3014553014553014,
      "grad_norm": 4.880962371826172,
      "learning_rate": 3.7768537768537773e-06,
      "loss": 0.1439,
      "step": 3176
    },
    {
      "epoch": 3.3024948024948024,
      "grad_norm": 10.522567749023438,
      "learning_rate": 3.7745437745437746e-06,
      "loss": 0.6237,
      "step": 3177
    },
    {
      "epoch": 3.3035343035343034,
      "grad_norm": 3.7333545684814453,
      "learning_rate": 3.772233772233773e-06,
      "loss": 0.0962,
      "step": 3178
    },
    {
      "epoch": 3.3045738045738045,
      "grad_norm": 0.20931552350521088,
      "learning_rate": 3.76992376992377e-06,
      "loss": 0.0057,
      "step": 3179
    },
    {
      "epoch": 3.3056133056133055,
      "grad_norm": 0.03920707851648331,
      "learning_rate": 3.767613767613768e-06,
      "loss": 0.0009,
      "step": 3180
    },
    {
      "epoch": 3.3066528066528065,
      "grad_norm": 0.022222192957997322,
      "learning_rate": 3.7653037653037657e-06,
      "loss": 0.0007,
      "step": 3181
    },
    {
      "epoch": 3.3076923076923075,
      "grad_norm": 10.037214279174805,
      "learning_rate": 3.7629937629937634e-06,
      "loss": 0.6276,
      "step": 3182
    },
    {
      "epoch": 3.3087318087318085,
      "grad_norm": 0.10803408175706863,
      "learning_rate": 3.760683760683761e-06,
      "loss": 0.0022,
      "step": 3183
    },
    {
      "epoch": 3.30977130977131,
      "grad_norm": 3.0191409587860107,
      "learning_rate": 3.758373758373759e-06,
      "loss": 0.05,
      "step": 3184
    },
    {
      "epoch": 3.310810810810811,
      "grad_norm": 0.004974550101906061,
      "learning_rate": 3.7560637560637563e-06,
      "loss": 0.0001,
      "step": 3185
    },
    {
      "epoch": 3.311850311850312,
      "grad_norm": 2.488440990447998,
      "learning_rate": 3.7537537537537537e-06,
      "loss": 0.0761,
      "step": 3186
    },
    {
      "epoch": 3.312889812889813,
      "grad_norm": 0.1313776671886444,
      "learning_rate": 3.751443751443752e-06,
      "loss": 0.0036,
      "step": 3187
    },
    {
      "epoch": 3.313929313929314,
      "grad_norm": 0.7895636558532715,
      "learning_rate": 3.749133749133749e-06,
      "loss": 0.0161,
      "step": 3188
    },
    {
      "epoch": 3.314968814968815,
      "grad_norm": 0.1644628494977951,
      "learning_rate": 3.7468237468237474e-06,
      "loss": 0.0032,
      "step": 3189
    },
    {
      "epoch": 3.316008316008316,
      "grad_norm": 7.237555980682373,
      "learning_rate": 3.7445137445137447e-06,
      "loss": 0.2825,
      "step": 3190
    },
    {
      "epoch": 3.317047817047817,
      "grad_norm": 0.4008646607398987,
      "learning_rate": 3.7422037422037425e-06,
      "loss": 0.0082,
      "step": 3191
    },
    {
      "epoch": 3.318087318087318,
      "grad_norm": 0.002861259737983346,
      "learning_rate": 3.7398937398937403e-06,
      "loss": 0.0001,
      "step": 3192
    },
    {
      "epoch": 3.319126819126819,
      "grad_norm": 0.028044261038303375,
      "learning_rate": 3.737583737583738e-06,
      "loss": 0.0005,
      "step": 3193
    },
    {
      "epoch": 3.32016632016632,
      "grad_norm": 0.0059648011811077595,
      "learning_rate": 3.735273735273736e-06,
      "loss": 0.0002,
      "step": 3194
    },
    {
      "epoch": 3.321205821205821,
      "grad_norm": 0.5465347766876221,
      "learning_rate": 3.732963732963733e-06,
      "loss": 0.0163,
      "step": 3195
    },
    {
      "epoch": 3.322245322245322,
      "grad_norm": 1.3572473526000977,
      "learning_rate": 3.730653730653731e-06,
      "loss": 0.0245,
      "step": 3196
    },
    {
      "epoch": 3.323284823284823,
      "grad_norm": 10.709805488586426,
      "learning_rate": 3.7283437283437287e-06,
      "loss": 1.354,
      "step": 3197
    },
    {
      "epoch": 3.3243243243243246,
      "grad_norm": 8.527897834777832,
      "learning_rate": 3.7260337260337265e-06,
      "loss": 0.2518,
      "step": 3198
    },
    {
      "epoch": 3.3253638253638256,
      "grad_norm": 0.12114723026752472,
      "learning_rate": 3.723723723723724e-06,
      "loss": 0.0044,
      "step": 3199
    },
    {
      "epoch": 3.3264033264033266,
      "grad_norm": 1.078540563583374,
      "learning_rate": 3.721413721413722e-06,
      "loss": 0.0328,
      "step": 3200
    },
    {
      "epoch": 3.3274428274428276,
      "grad_norm": 0.03818194568157196,
      "learning_rate": 3.7191037191037193e-06,
      "loss": 0.001,
      "step": 3201
    },
    {
      "epoch": 3.3284823284823286,
      "grad_norm": 9.92434024810791,
      "learning_rate": 3.7167937167937175e-06,
      "loss": 0.4603,
      "step": 3202
    },
    {
      "epoch": 3.3295218295218296,
      "grad_norm": 7.579704761505127,
      "learning_rate": 3.714483714483715e-06,
      "loss": 0.3058,
      "step": 3203
    },
    {
      "epoch": 3.3305613305613306,
      "grad_norm": 10.627359390258789,
      "learning_rate": 3.712173712173712e-06,
      "loss": 1.1327,
      "step": 3204
    },
    {
      "epoch": 3.3316008316008316,
      "grad_norm": 11.167915344238281,
      "learning_rate": 3.7098637098637104e-06,
      "loss": 0.4044,
      "step": 3205
    },
    {
      "epoch": 3.3326403326403327,
      "grad_norm": 9.741639137268066,
      "learning_rate": 3.7075537075537077e-06,
      "loss": 0.5733,
      "step": 3206
    },
    {
      "epoch": 3.3336798336798337,
      "grad_norm": 21.80474090576172,
      "learning_rate": 3.7052437052437055e-06,
      "loss": 1.0578,
      "step": 3207
    },
    {
      "epoch": 3.3347193347193347,
      "grad_norm": 11.758549690246582,
      "learning_rate": 3.7029337029337033e-06,
      "loss": 0.8109,
      "step": 3208
    },
    {
      "epoch": 3.3357588357588357,
      "grad_norm": 2.411346197128296,
      "learning_rate": 3.700623700623701e-06,
      "loss": 0.0536,
      "step": 3209
    },
    {
      "epoch": 3.3367983367983367,
      "grad_norm": 0.29183125495910645,
      "learning_rate": 3.6983136983136984e-06,
      "loss": 0.0056,
      "step": 3210
    },
    {
      "epoch": 3.3378378378378377,
      "grad_norm": 8.743294715881348,
      "learning_rate": 3.6960036960036966e-06,
      "loss": 0.9973,
      "step": 3211
    },
    {
      "epoch": 3.3388773388773387,
      "grad_norm": 4.9171953201293945,
      "learning_rate": 3.693693693693694e-06,
      "loss": 0.1239,
      "step": 3212
    },
    {
      "epoch": 3.3399168399168397,
      "grad_norm": 9.574235916137695,
      "learning_rate": 3.6913836913836913e-06,
      "loss": 0.245,
      "step": 3213
    },
    {
      "epoch": 3.3409563409563408,
      "grad_norm": 10.361645698547363,
      "learning_rate": 3.6890736890736895e-06,
      "loss": 0.5266,
      "step": 3214
    },
    {
      "epoch": 3.3419958419958418,
      "grad_norm": 1.520453691482544,
      "learning_rate": 3.686763686763687e-06,
      "loss": 0.2928,
      "step": 3215
    },
    {
      "epoch": 3.343035343035343,
      "grad_norm": 1.8686317205429077,
      "learning_rate": 3.684453684453685e-06,
      "loss": 0.0391,
      "step": 3216
    },
    {
      "epoch": 3.3440748440748442,
      "grad_norm": 8.648658752441406,
      "learning_rate": 3.6821436821436823e-06,
      "loss": 1.0406,
      "step": 3217
    },
    {
      "epoch": 3.3451143451143452,
      "grad_norm": 2.9640727043151855,
      "learning_rate": 3.6798336798336805e-06,
      "loss": 0.0632,
      "step": 3218
    },
    {
      "epoch": 3.3461538461538463,
      "grad_norm": 2.495372772216797,
      "learning_rate": 3.677523677523678e-06,
      "loss": 0.0467,
      "step": 3219
    },
    {
      "epoch": 3.3471933471933473,
      "grad_norm": 10.242700576782227,
      "learning_rate": 3.6752136752136756e-06,
      "loss": 0.7899,
      "step": 3220
    },
    {
      "epoch": 3.3482328482328483,
      "grad_norm": 5.957021713256836,
      "learning_rate": 3.6729036729036734e-06,
      "loss": 0.7897,
      "step": 3221
    },
    {
      "epoch": 3.3492723492723493,
      "grad_norm": 0.18006573617458344,
      "learning_rate": 3.6705936705936707e-06,
      "loss": 0.0059,
      "step": 3222
    },
    {
      "epoch": 3.3503118503118503,
      "grad_norm": 0.0461384654045105,
      "learning_rate": 3.6682836682836685e-06,
      "loss": 0.0013,
      "step": 3223
    },
    {
      "epoch": 3.3513513513513513,
      "grad_norm": 6.071020603179932,
      "learning_rate": 3.6659736659736663e-06,
      "loss": 0.2303,
      "step": 3224
    },
    {
      "epoch": 3.3523908523908523,
      "grad_norm": 1.346615195274353,
      "learning_rate": 3.663663663663664e-06,
      "loss": 0.0333,
      "step": 3225
    },
    {
      "epoch": 3.3534303534303533,
      "grad_norm": 0.11079871654510498,
      "learning_rate": 3.6613536613536614e-06,
      "loss": 0.0038,
      "step": 3226
    },
    {
      "epoch": 3.3544698544698544,
      "grad_norm": 3.9040043354034424,
      "learning_rate": 3.6590436590436596e-06,
      "loss": 0.1225,
      "step": 3227
    },
    {
      "epoch": 3.3555093555093554,
      "grad_norm": 6.001784324645996,
      "learning_rate": 3.656733656733657e-06,
      "loss": 0.3533,
      "step": 3228
    },
    {
      "epoch": 3.3565488565488564,
      "grad_norm": 6.267105579376221,
      "learning_rate": 3.654423654423655e-06,
      "loss": 0.189,
      "step": 3229
    },
    {
      "epoch": 3.357588357588358,
      "grad_norm": 9.05850601196289,
      "learning_rate": 3.6521136521136525e-06,
      "loss": 1.8204,
      "step": 3230
    },
    {
      "epoch": 3.358627858627859,
      "grad_norm": 5.095354080200195,
      "learning_rate": 3.64980364980365e-06,
      "loss": 0.1764,
      "step": 3231
    },
    {
      "epoch": 3.35966735966736,
      "grad_norm": 0.06016012281179428,
      "learning_rate": 3.647493647493648e-06,
      "loss": 0.0014,
      "step": 3232
    },
    {
      "epoch": 3.360706860706861,
      "grad_norm": 4.24834680557251,
      "learning_rate": 3.6451836451836453e-06,
      "loss": 0.2245,
      "step": 3233
    },
    {
      "epoch": 3.361746361746362,
      "grad_norm": 0.030917443335056305,
      "learning_rate": 3.642873642873643e-06,
      "loss": 0.0007,
      "step": 3234
    },
    {
      "epoch": 3.362785862785863,
      "grad_norm": 0.8087771534919739,
      "learning_rate": 3.640563640563641e-06,
      "loss": 0.0179,
      "step": 3235
    },
    {
      "epoch": 3.363825363825364,
      "grad_norm": 0.3363551199436188,
      "learning_rate": 3.6382536382536386e-06,
      "loss": 0.0066,
      "step": 3236
    },
    {
      "epoch": 3.364864864864865,
      "grad_norm": 8.794129371643066,
      "learning_rate": 3.635943635943636e-06,
      "loss": 0.6676,
      "step": 3237
    },
    {
      "epoch": 3.365904365904366,
      "grad_norm": 5.283533096313477,
      "learning_rate": 3.633633633633634e-06,
      "loss": 0.1615,
      "step": 3238
    },
    {
      "epoch": 3.366943866943867,
      "grad_norm": 0.8963358402252197,
      "learning_rate": 3.6313236313236315e-06,
      "loss": 0.0214,
      "step": 3239
    },
    {
      "epoch": 3.367983367983368,
      "grad_norm": 1.7845673561096191,
      "learning_rate": 3.6290136290136293e-06,
      "loss": 0.0336,
      "step": 3240
    },
    {
      "epoch": 3.369022869022869,
      "grad_norm": 3.5747463703155518,
      "learning_rate": 3.626703626703627e-06,
      "loss": 0.0929,
      "step": 3241
    },
    {
      "epoch": 3.37006237006237,
      "grad_norm": 1.0860050916671753,
      "learning_rate": 3.6243936243936244e-06,
      "loss": 0.0163,
      "step": 3242
    },
    {
      "epoch": 3.371101871101871,
      "grad_norm": 2.0682332515716553,
      "learning_rate": 3.6220836220836226e-06,
      "loss": 0.0467,
      "step": 3243
    },
    {
      "epoch": 3.372141372141372,
      "grad_norm": 0.3948007822036743,
      "learning_rate": 3.61977361977362e-06,
      "loss": 0.01,
      "step": 3244
    },
    {
      "epoch": 3.373180873180873,
      "grad_norm": 0.7122275233268738,
      "learning_rate": 3.617463617463618e-06,
      "loss": 0.0148,
      "step": 3245
    },
    {
      "epoch": 3.374220374220374,
      "grad_norm": 3.147752046585083,
      "learning_rate": 3.6151536151536155e-06,
      "loss": 0.1012,
      "step": 3246
    },
    {
      "epoch": 3.375259875259875,
      "grad_norm": 8.694098472595215,
      "learning_rate": 3.6128436128436132e-06,
      "loss": 0.3972,
      "step": 3247
    },
    {
      "epoch": 3.3762993762993765,
      "grad_norm": 6.065917491912842,
      "learning_rate": 3.610533610533611e-06,
      "loss": 0.1205,
      "step": 3248
    },
    {
      "epoch": 3.3773388773388775,
      "grad_norm": 0.15893030166625977,
      "learning_rate": 3.6082236082236083e-06,
      "loss": 0.0049,
      "step": 3249
    },
    {
      "epoch": 3.3783783783783785,
      "grad_norm": 1.9620164632797241,
      "learning_rate": 3.605913605913606e-06,
      "loss": 0.0327,
      "step": 3250
    },
    {
      "epoch": 3.3794178794178795,
      "grad_norm": 6.791231155395508,
      "learning_rate": 3.603603603603604e-06,
      "loss": 0.724,
      "step": 3251
    },
    {
      "epoch": 3.3804573804573805,
      "grad_norm": 1.8762073516845703,
      "learning_rate": 3.6012936012936016e-06,
      "loss": 0.0796,
      "step": 3252
    },
    {
      "epoch": 3.3814968814968815,
      "grad_norm": 2.9928243160247803,
      "learning_rate": 3.598983598983599e-06,
      "loss": 0.2971,
      "step": 3253
    },
    {
      "epoch": 3.3825363825363826,
      "grad_norm": 1.9753413200378418,
      "learning_rate": 3.596673596673597e-06,
      "loss": 0.0759,
      "step": 3254
    },
    {
      "epoch": 3.3835758835758836,
      "grad_norm": 5.722926616668701,
      "learning_rate": 3.5943635943635945e-06,
      "loss": 0.1421,
      "step": 3255
    },
    {
      "epoch": 3.3846153846153846,
      "grad_norm": 6.9614481925964355,
      "learning_rate": 3.5920535920535927e-06,
      "loss": 0.2756,
      "step": 3256
    },
    {
      "epoch": 3.3856548856548856,
      "grad_norm": 1.7708218097686768,
      "learning_rate": 3.58974358974359e-06,
      "loss": 0.0287,
      "step": 3257
    },
    {
      "epoch": 3.3866943866943866,
      "grad_norm": 4.310425758361816,
      "learning_rate": 3.5874335874335874e-06,
      "loss": 0.3408,
      "step": 3258
    },
    {
      "epoch": 3.3877338877338876,
      "grad_norm": 0.33174580335617065,
      "learning_rate": 3.5851235851235856e-06,
      "loss": 0.0075,
      "step": 3259
    },
    {
      "epoch": 3.3887733887733886,
      "grad_norm": 0.6925557851791382,
      "learning_rate": 3.582813582813583e-06,
      "loss": 0.018,
      "step": 3260
    },
    {
      "epoch": 3.3898128898128896,
      "grad_norm": 5.32797908782959,
      "learning_rate": 3.580503580503581e-06,
      "loss": 0.6188,
      "step": 3261
    },
    {
      "epoch": 3.390852390852391,
      "grad_norm": 13.79472541809082,
      "learning_rate": 3.5781935781935785e-06,
      "loss": 1.1053,
      "step": 3262
    },
    {
      "epoch": 3.391891891891892,
      "grad_norm": 4.517867088317871,
      "learning_rate": 3.5758835758835762e-06,
      "loss": 0.159,
      "step": 3263
    },
    {
      "epoch": 3.392931392931393,
      "grad_norm": 10.487228393554688,
      "learning_rate": 3.573573573573574e-06,
      "loss": 0.6029,
      "step": 3264
    },
    {
      "epoch": 3.393970893970894,
      "grad_norm": 0.06305764615535736,
      "learning_rate": 3.5712635712635718e-06,
      "loss": 0.0014,
      "step": 3265
    },
    {
      "epoch": 3.395010395010395,
      "grad_norm": 0.005865427199751139,
      "learning_rate": 3.568953568953569e-06,
      "loss": 0.0001,
      "step": 3266
    },
    {
      "epoch": 3.396049896049896,
      "grad_norm": 0.030989723280072212,
      "learning_rate": 3.566643566643567e-06,
      "loss": 0.0006,
      "step": 3267
    },
    {
      "epoch": 3.397089397089397,
      "grad_norm": 5.602653980255127,
      "learning_rate": 3.5643335643335647e-06,
      "loss": 0.0412,
      "step": 3268
    },
    {
      "epoch": 3.398128898128898,
      "grad_norm": 17.178709030151367,
      "learning_rate": 3.562023562023562e-06,
      "loss": 1.5057,
      "step": 3269
    },
    {
      "epoch": 3.399168399168399,
      "grad_norm": 0.24581624567508698,
      "learning_rate": 3.55971355971356e-06,
      "loss": 0.0068,
      "step": 3270
    },
    {
      "epoch": 3.4002079002079,
      "grad_norm": 3.708796501159668,
      "learning_rate": 3.5574035574035575e-06,
      "loss": 0.0858,
      "step": 3271
    },
    {
      "epoch": 3.401247401247401,
      "grad_norm": 5.2083868980407715,
      "learning_rate": 3.5550935550935557e-06,
      "loss": 0.124,
      "step": 3272
    },
    {
      "epoch": 3.4022869022869022,
      "grad_norm": 0.6852853894233704,
      "learning_rate": 3.552783552783553e-06,
      "loss": 0.0206,
      "step": 3273
    },
    {
      "epoch": 3.4033264033264032,
      "grad_norm": 7.372675895690918,
      "learning_rate": 3.550473550473551e-06,
      "loss": 0.1647,
      "step": 3274
    },
    {
      "epoch": 3.4043659043659042,
      "grad_norm": 7.724948406219482,
      "learning_rate": 3.5481635481635486e-06,
      "loss": 0.2526,
      "step": 3275
    },
    {
      "epoch": 3.4054054054054053,
      "grad_norm": 8.5491304397583,
      "learning_rate": 3.545853545853546e-06,
      "loss": 0.7543,
      "step": 3276
    },
    {
      "epoch": 3.4064449064449063,
      "grad_norm": 1.5162405967712402,
      "learning_rate": 3.5435435435435437e-06,
      "loss": 0.0232,
      "step": 3277
    },
    {
      "epoch": 3.4074844074844073,
      "grad_norm": 6.716646194458008,
      "learning_rate": 3.5412335412335415e-06,
      "loss": 0.1469,
      "step": 3278
    },
    {
      "epoch": 3.4085239085239083,
      "grad_norm": 1.3024156093597412,
      "learning_rate": 3.5389235389235392e-06,
      "loss": 0.0274,
      "step": 3279
    },
    {
      "epoch": 3.4095634095634098,
      "grad_norm": 2.4426019191741943,
      "learning_rate": 3.5366135366135366e-06,
      "loss": 0.0742,
      "step": 3280
    },
    {
      "epoch": 3.4106029106029108,
      "grad_norm": 1.7469676733016968,
      "learning_rate": 3.5343035343035348e-06,
      "loss": 0.0559,
      "step": 3281
    },
    {
      "epoch": 3.4116424116424118,
      "grad_norm": 0.047953590750694275,
      "learning_rate": 3.531993531993532e-06,
      "loss": 0.001,
      "step": 3282
    },
    {
      "epoch": 3.412681912681913,
      "grad_norm": 9.703154563903809,
      "learning_rate": 3.5296835296835303e-06,
      "loss": 0.39,
      "step": 3283
    },
    {
      "epoch": 3.413721413721414,
      "grad_norm": 2.85369873046875,
      "learning_rate": 3.5273735273735277e-06,
      "loss": 0.0528,
      "step": 3284
    },
    {
      "epoch": 3.414760914760915,
      "grad_norm": 10.792146682739258,
      "learning_rate": 3.525063525063525e-06,
      "loss": 0.5995,
      "step": 3285
    },
    {
      "epoch": 3.415800415800416,
      "grad_norm": 7.268505096435547,
      "learning_rate": 3.522753522753523e-06,
      "loss": 0.2875,
      "step": 3286
    },
    {
      "epoch": 3.416839916839917,
      "grad_norm": 0.2587016224861145,
      "learning_rate": 3.5204435204435205e-06,
      "loss": 0.0075,
      "step": 3287
    },
    {
      "epoch": 3.417879417879418,
      "grad_norm": 4.772622108459473,
      "learning_rate": 3.5181335181335187e-06,
      "loss": 0.4761,
      "step": 3288
    },
    {
      "epoch": 3.418918918918919,
      "grad_norm": 8.32681941986084,
      "learning_rate": 3.515823515823516e-06,
      "loss": 0.2824,
      "step": 3289
    },
    {
      "epoch": 3.41995841995842,
      "grad_norm": 8.72518253326416,
      "learning_rate": 3.513513513513514e-06,
      "loss": 0.5085,
      "step": 3290
    },
    {
      "epoch": 3.420997920997921,
      "grad_norm": 0.007968052290380001,
      "learning_rate": 3.5112035112035116e-06,
      "loss": 0.0002,
      "step": 3291
    },
    {
      "epoch": 3.422037422037422,
      "grad_norm": 14.061298370361328,
      "learning_rate": 3.5088935088935094e-06,
      "loss": 1.0375,
      "step": 3292
    },
    {
      "epoch": 3.423076923076923,
      "grad_norm": 6.244088649749756,
      "learning_rate": 3.5065835065835067e-06,
      "loss": 0.1486,
      "step": 3293
    },
    {
      "epoch": 3.4241164241164244,
      "grad_norm": 9.241902351379395,
      "learning_rate": 3.5042735042735045e-06,
      "loss": 0.7761,
      "step": 3294
    },
    {
      "epoch": 3.4251559251559254,
      "grad_norm": 7.88060188293457,
      "learning_rate": 3.5019635019635022e-06,
      "loss": 0.701,
      "step": 3295
    },
    {
      "epoch": 3.4261954261954264,
      "grad_norm": 10.947381973266602,
      "learning_rate": 3.4996534996534996e-06,
      "loss": 0.8914,
      "step": 3296
    },
    {
      "epoch": 3.4272349272349274,
      "grad_norm": 8.519932746887207,
      "learning_rate": 3.4973434973434978e-06,
      "loss": 0.6309,
      "step": 3297
    },
    {
      "epoch": 3.4282744282744284,
      "grad_norm": 0.453152596950531,
      "learning_rate": 3.495033495033495e-06,
      "loss": 0.0069,
      "step": 3298
    },
    {
      "epoch": 3.4293139293139294,
      "grad_norm": 5.153072834014893,
      "learning_rate": 3.4927234927234933e-06,
      "loss": 0.3492,
      "step": 3299
    },
    {
      "epoch": 3.4303534303534304,
      "grad_norm": 0.113200344145298,
      "learning_rate": 3.4904134904134907e-06,
      "loss": 0.0024,
      "step": 3300
    },
    {
      "epoch": 3.4313929313929314,
      "grad_norm": 0.620791494846344,
      "learning_rate": 3.488103488103489e-06,
      "loss": 0.009,
      "step": 3301
    },
    {
      "epoch": 3.4324324324324325,
      "grad_norm": 9.207609176635742,
      "learning_rate": 3.485793485793486e-06,
      "loss": 0.6816,
      "step": 3302
    },
    {
      "epoch": 3.4334719334719335,
      "grad_norm": 7.150522232055664,
      "learning_rate": 3.4834834834834835e-06,
      "loss": 0.3006,
      "step": 3303
    },
    {
      "epoch": 3.4345114345114345,
      "grad_norm": 14.311772346496582,
      "learning_rate": 3.4811734811734817e-06,
      "loss": 0.6698,
      "step": 3304
    },
    {
      "epoch": 3.4355509355509355,
      "grad_norm": 5.834444522857666,
      "learning_rate": 3.478863478863479e-06,
      "loss": 0.2432,
      "step": 3305
    },
    {
      "epoch": 3.4365904365904365,
      "grad_norm": 5.6178388595581055,
      "learning_rate": 3.476553476553477e-06,
      "loss": 0.2244,
      "step": 3306
    },
    {
      "epoch": 3.4376299376299375,
      "grad_norm": 8.573112487792969,
      "learning_rate": 3.4742434742434746e-06,
      "loss": 0.2221,
      "step": 3307
    },
    {
      "epoch": 3.4386694386694385,
      "grad_norm": 1.456484317779541,
      "learning_rate": 3.4719334719334724e-06,
      "loss": 0.027,
      "step": 3308
    },
    {
      "epoch": 3.4397089397089395,
      "grad_norm": 8.12042236328125,
      "learning_rate": 3.4696234696234697e-06,
      "loss": 0.1341,
      "step": 3309
    },
    {
      "epoch": 3.4407484407484406,
      "grad_norm": 2.8493335247039795,
      "learning_rate": 3.467313467313468e-06,
      "loss": 0.0507,
      "step": 3310
    },
    {
      "epoch": 3.4417879417879416,
      "grad_norm": 0.18166130781173706,
      "learning_rate": 3.4650034650034653e-06,
      "loss": 0.0043,
      "step": 3311
    },
    {
      "epoch": 3.442827442827443,
      "grad_norm": 0.8375445008277893,
      "learning_rate": 3.4626934626934626e-06,
      "loss": 0.0124,
      "step": 3312
    },
    {
      "epoch": 3.443866943866944,
      "grad_norm": 3.7751896381378174,
      "learning_rate": 3.4603834603834608e-06,
      "loss": 0.2906,
      "step": 3313
    },
    {
      "epoch": 3.444906444906445,
      "grad_norm": 6.509339809417725,
      "learning_rate": 3.458073458073458e-06,
      "loss": 0.3631,
      "step": 3314
    },
    {
      "epoch": 3.445945945945946,
      "grad_norm": 8.254549026489258,
      "learning_rate": 3.4557634557634563e-06,
      "loss": 0.2388,
      "step": 3315
    },
    {
      "epoch": 3.446985446985447,
      "grad_norm": 7.653320789337158,
      "learning_rate": 3.4534534534534537e-06,
      "loss": 0.2905,
      "step": 3316
    },
    {
      "epoch": 3.448024948024948,
      "grad_norm": 13.026524543762207,
      "learning_rate": 3.4511434511434514e-06,
      "loss": 0.4787,
      "step": 3317
    },
    {
      "epoch": 3.449064449064449,
      "grad_norm": 3.805297374725342,
      "learning_rate": 3.448833448833449e-06,
      "loss": 0.1155,
      "step": 3318
    },
    {
      "epoch": 3.45010395010395,
      "grad_norm": 0.15272344648838043,
      "learning_rate": 3.446523446523447e-06,
      "loss": 0.0039,
      "step": 3319
    },
    {
      "epoch": 3.451143451143451,
      "grad_norm": 6.021086692810059,
      "learning_rate": 3.4442134442134443e-06,
      "loss": 0.3758,
      "step": 3320
    },
    {
      "epoch": 3.452182952182952,
      "grad_norm": 0.07570965588092804,
      "learning_rate": 3.441903441903442e-06,
      "loss": 0.0027,
      "step": 3321
    },
    {
      "epoch": 3.453222453222453,
      "grad_norm": 0.24904006719589233,
      "learning_rate": 3.43959343959344e-06,
      "loss": 0.0048,
      "step": 3322
    },
    {
      "epoch": 3.454261954261954,
      "grad_norm": 21.89639663696289,
      "learning_rate": 3.437283437283437e-06,
      "loss": 0.4499,
      "step": 3323
    },
    {
      "epoch": 3.455301455301455,
      "grad_norm": 0.08154599368572235,
      "learning_rate": 3.4349734349734354e-06,
      "loss": 0.0018,
      "step": 3324
    },
    {
      "epoch": 3.456340956340956,
      "grad_norm": 0.4847028851509094,
      "learning_rate": 3.4326634326634327e-06,
      "loss": 0.0132,
      "step": 3325
    },
    {
      "epoch": 3.4573804573804576,
      "grad_norm": 13.547603607177734,
      "learning_rate": 3.430353430353431e-06,
      "loss": 1.0378,
      "step": 3326
    },
    {
      "epoch": 3.4584199584199586,
      "grad_norm": 10.268739700317383,
      "learning_rate": 3.4280434280434283e-06,
      "loss": 0.6069,
      "step": 3327
    },
    {
      "epoch": 3.4594594594594597,
      "grad_norm": 0.10986010730266571,
      "learning_rate": 3.4257334257334264e-06,
      "loss": 0.0018,
      "step": 3328
    },
    {
      "epoch": 3.4604989604989607,
      "grad_norm": 1.7400387525558472,
      "learning_rate": 3.423423423423424e-06,
      "loss": 0.0485,
      "step": 3329
    },
    {
      "epoch": 3.4615384615384617,
      "grad_norm": 3.423602819442749,
      "learning_rate": 3.421113421113421e-06,
      "loss": 0.1067,
      "step": 3330
    },
    {
      "epoch": 3.4625779625779627,
      "grad_norm": 0.472627729177475,
      "learning_rate": 3.4188034188034193e-06,
      "loss": 0.0154,
      "step": 3331
    },
    {
      "epoch": 3.4636174636174637,
      "grad_norm": 0.05725175887346268,
      "learning_rate": 3.4164934164934167e-06,
      "loss": 0.0017,
      "step": 3332
    },
    {
      "epoch": 3.4646569646569647,
      "grad_norm": 0.6359158754348755,
      "learning_rate": 3.4141834141834144e-06,
      "loss": 0.0094,
      "step": 3333
    },
    {
      "epoch": 3.4656964656964657,
      "grad_norm": 7.547875881195068,
      "learning_rate": 3.411873411873412e-06,
      "loss": 0.5435,
      "step": 3334
    },
    {
      "epoch": 3.4667359667359667,
      "grad_norm": 0.7937979698181152,
      "learning_rate": 3.40956340956341e-06,
      "loss": 0.0177,
      "step": 3335
    },
    {
      "epoch": 3.4677754677754677,
      "grad_norm": 10.875249862670898,
      "learning_rate": 3.4072534072534073e-06,
      "loss": 0.9046,
      "step": 3336
    },
    {
      "epoch": 3.4688149688149688,
      "grad_norm": 0.3446725606918335,
      "learning_rate": 3.4049434049434055e-06,
      "loss": 0.0061,
      "step": 3337
    },
    {
      "epoch": 3.4698544698544698,
      "grad_norm": 4.590447425842285,
      "learning_rate": 3.402633402633403e-06,
      "loss": 0.1371,
      "step": 3338
    },
    {
      "epoch": 3.470893970893971,
      "grad_norm": 6.492908954620361,
      "learning_rate": 3.4003234003234e-06,
      "loss": 0.3914,
      "step": 3339
    },
    {
      "epoch": 3.471933471933472,
      "grad_norm": 0.5407949686050415,
      "learning_rate": 3.3980133980133984e-06,
      "loss": 0.0078,
      "step": 3340
    },
    {
      "epoch": 3.472972972972973,
      "grad_norm": 14.164336204528809,
      "learning_rate": 3.3957033957033957e-06,
      "loss": 1.6147,
      "step": 3341
    },
    {
      "epoch": 3.474012474012474,
      "grad_norm": 3.5835986137390137,
      "learning_rate": 3.393393393393394e-06,
      "loss": 0.0709,
      "step": 3342
    },
    {
      "epoch": 3.475051975051975,
      "grad_norm": 0.34511393308639526,
      "learning_rate": 3.3910833910833913e-06,
      "loss": 0.008,
      "step": 3343
    },
    {
      "epoch": 3.4760914760914763,
      "grad_norm": 2.1724395751953125,
      "learning_rate": 3.388773388773389e-06,
      "loss": 0.0773,
      "step": 3344
    },
    {
      "epoch": 3.4771309771309773,
      "grad_norm": 0.025691157206892967,
      "learning_rate": 3.386463386463387e-06,
      "loss": 0.0007,
      "step": 3345
    },
    {
      "epoch": 3.4781704781704783,
      "grad_norm": 0.0027891502249985933,
      "learning_rate": 3.3841533841533846e-06,
      "loss": 0.0001,
      "step": 3346
    },
    {
      "epoch": 3.4792099792099793,
      "grad_norm": 7.376180171966553,
      "learning_rate": 3.3818433818433823e-06,
      "loss": 0.7187,
      "step": 3347
    },
    {
      "epoch": 3.4802494802494803,
      "grad_norm": 0.02257869578897953,
      "learning_rate": 3.3795333795333797e-06,
      "loss": 0.0006,
      "step": 3348
    },
    {
      "epoch": 3.4812889812889813,
      "grad_norm": 10.220701217651367,
      "learning_rate": 3.3772233772233774e-06,
      "loss": 0.5355,
      "step": 3349
    },
    {
      "epoch": 3.4823284823284824,
      "grad_norm": 3.1495954990386963,
      "learning_rate": 3.374913374913375e-06,
      "loss": 0.3358,
      "step": 3350
    },
    {
      "epoch": 3.4833679833679834,
      "grad_norm": 5.3782830238342285,
      "learning_rate": 3.372603372603373e-06,
      "loss": 0.1578,
      "step": 3351
    },
    {
      "epoch": 3.4844074844074844,
      "grad_norm": 0.06863047927618027,
      "learning_rate": 3.3702933702933703e-06,
      "loss": 0.0012,
      "step": 3352
    },
    {
      "epoch": 3.4854469854469854,
      "grad_norm": 1.133750319480896,
      "learning_rate": 3.3679833679833685e-06,
      "loss": 0.0422,
      "step": 3353
    },
    {
      "epoch": 3.4864864864864864,
      "grad_norm": 11.342142105102539,
      "learning_rate": 3.365673365673366e-06,
      "loss": 1.5505,
      "step": 3354
    },
    {
      "epoch": 3.4875259875259874,
      "grad_norm": 1.8219274282455444,
      "learning_rate": 3.363363363363364e-06,
      "loss": 0.0329,
      "step": 3355
    },
    {
      "epoch": 3.4885654885654884,
      "grad_norm": 5.730031967163086,
      "learning_rate": 3.3610533610533614e-06,
      "loss": 0.2092,
      "step": 3356
    },
    {
      "epoch": 3.4896049896049894,
      "grad_norm": 0.18049009144306183,
      "learning_rate": 3.3587433587433587e-06,
      "loss": 0.0041,
      "step": 3357
    },
    {
      "epoch": 3.490644490644491,
      "grad_norm": 2.5031416416168213,
      "learning_rate": 3.356433356433357e-06,
      "loss": 0.0972,
      "step": 3358
    },
    {
      "epoch": 3.491683991683992,
      "grad_norm": 0.024312352761626244,
      "learning_rate": 3.3541233541233543e-06,
      "loss": 0.0007,
      "step": 3359
    },
    {
      "epoch": 3.492723492723493,
      "grad_norm": 1.044015884399414,
      "learning_rate": 3.351813351813352e-06,
      "loss": 0.0218,
      "step": 3360
    },
    {
      "epoch": 3.493762993762994,
      "grad_norm": 9.544400215148926,
      "learning_rate": 3.34950334950335e-06,
      "loss": 0.8989,
      "step": 3361
    },
    {
      "epoch": 3.494802494802495,
      "grad_norm": 3.1391563415527344,
      "learning_rate": 3.3471933471933476e-06,
      "loss": 0.0732,
      "step": 3362
    },
    {
      "epoch": 3.495841995841996,
      "grad_norm": 0.5424788594245911,
      "learning_rate": 3.344883344883345e-06,
      "loss": 0.0115,
      "step": 3363
    },
    {
      "epoch": 3.496881496881497,
      "grad_norm": 11.996427536010742,
      "learning_rate": 3.342573342573343e-06,
      "loss": 0.8597,
      "step": 3364
    },
    {
      "epoch": 3.497920997920998,
      "grad_norm": 0.0049547902308404446,
      "learning_rate": 3.3402633402633404e-06,
      "loss": 0.0001,
      "step": 3365
    },
    {
      "epoch": 3.498960498960499,
      "grad_norm": 0.07860179245471954,
      "learning_rate": 3.337953337953338e-06,
      "loss": 0.0013,
      "step": 3366
    },
    {
      "epoch": 3.5,
      "grad_norm": 10.51418685913086,
      "learning_rate": 3.335643335643336e-06,
      "loss": 0.2187,
      "step": 3367
    },
    {
      "epoch": 3.501039501039501,
      "grad_norm": 0.04997267574071884,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0013,
      "step": 3368
    },
    {
      "epoch": 3.502079002079002,
      "grad_norm": 11.178793907165527,
      "learning_rate": 3.3310233310233315e-06,
      "loss": 0.9946,
      "step": 3369
    },
    {
      "epoch": 3.503118503118503,
      "grad_norm": 4.278314113616943,
      "learning_rate": 3.328713328713329e-06,
      "loss": 0.1087,
      "step": 3370
    },
    {
      "epoch": 3.504158004158004,
      "grad_norm": 1.7856881618499756,
      "learning_rate": 3.326403326403327e-06,
      "loss": 0.0257,
      "step": 3371
    },
    {
      "epoch": 3.505197505197505,
      "grad_norm": 0.048529352992773056,
      "learning_rate": 3.3240933240933244e-06,
      "loss": 0.0016,
      "step": 3372
    },
    {
      "epoch": 3.506237006237006,
      "grad_norm": 3.9260010719299316,
      "learning_rate": 3.321783321783322e-06,
      "loss": 0.0494,
      "step": 3373
    },
    {
      "epoch": 3.507276507276507,
      "grad_norm": 3.7316524982452393,
      "learning_rate": 3.31947331947332e-06,
      "loss": 0.0984,
      "step": 3374
    },
    {
      "epoch": 3.508316008316008,
      "grad_norm": 7.0579400062561035,
      "learning_rate": 3.3171633171633173e-06,
      "loss": 0.2721,
      "step": 3375
    },
    {
      "epoch": 3.509355509355509,
      "grad_norm": 0.03118578903377056,
      "learning_rate": 3.314853314853315e-06,
      "loss": 0.0008,
      "step": 3376
    },
    {
      "epoch": 3.51039501039501,
      "grad_norm": 7.600438117980957,
      "learning_rate": 3.312543312543313e-06,
      "loss": 0.3741,
      "step": 3377
    },
    {
      "epoch": 3.5114345114345116,
      "grad_norm": 0.0009457019041292369,
      "learning_rate": 3.3102333102333106e-06,
      "loss": 0.0,
      "step": 3378
    },
    {
      "epoch": 3.5124740124740126,
      "grad_norm": 3.3967413902282715,
      "learning_rate": 3.307923307923308e-06,
      "loss": 0.08,
      "step": 3379
    },
    {
      "epoch": 3.5135135135135136,
      "grad_norm": 0.9884988069534302,
      "learning_rate": 3.305613305613306e-06,
      "loss": 0.0188,
      "step": 3380
    },
    {
      "epoch": 3.5145530145530146,
      "grad_norm": 0.1111251637339592,
      "learning_rate": 3.3033033033033035e-06,
      "loss": 0.0028,
      "step": 3381
    },
    {
      "epoch": 3.5155925155925156,
      "grad_norm": 9.056451797485352,
      "learning_rate": 3.3009933009933016e-06,
      "loss": 0.366,
      "step": 3382
    },
    {
      "epoch": 3.5166320166320166,
      "grad_norm": 5.030453681945801,
      "learning_rate": 3.298683298683299e-06,
      "loss": 0.1245,
      "step": 3383
    },
    {
      "epoch": 3.5176715176715176,
      "grad_norm": 9.545477867126465,
      "learning_rate": 3.2963732963732963e-06,
      "loss": 2.5062,
      "step": 3384
    },
    {
      "epoch": 3.5187110187110187,
      "grad_norm": 17.07159423828125,
      "learning_rate": 3.2940632940632945e-06,
      "loss": 0.78,
      "step": 3385
    },
    {
      "epoch": 3.5197505197505197,
      "grad_norm": 0.9238329529762268,
      "learning_rate": 3.291753291753292e-06,
      "loss": 0.0238,
      "step": 3386
    },
    {
      "epoch": 3.5207900207900207,
      "grad_norm": 3.1281144618988037,
      "learning_rate": 3.2894432894432896e-06,
      "loss": 0.2682,
      "step": 3387
    },
    {
      "epoch": 3.5218295218295217,
      "grad_norm": 7.184891700744629,
      "learning_rate": 3.2871332871332874e-06,
      "loss": 0.3436,
      "step": 3388
    },
    {
      "epoch": 3.5228690228690227,
      "grad_norm": 0.00028464957722462714,
      "learning_rate": 3.284823284823285e-06,
      "loss": 0.0,
      "step": 3389
    },
    {
      "epoch": 3.523908523908524,
      "grad_norm": 1.0748562812805176,
      "learning_rate": 3.2825132825132825e-06,
      "loss": 0.0262,
      "step": 3390
    },
    {
      "epoch": 3.524948024948025,
      "grad_norm": 0.48287853598594666,
      "learning_rate": 3.2802032802032807e-06,
      "loss": 0.0074,
      "step": 3391
    },
    {
      "epoch": 3.525987525987526,
      "grad_norm": 2.27337908744812,
      "learning_rate": 3.277893277893278e-06,
      "loss": 0.0436,
      "step": 3392
    },
    {
      "epoch": 3.527027027027027,
      "grad_norm": 13.80897045135498,
      "learning_rate": 3.275583275583276e-06,
      "loss": 0.3124,
      "step": 3393
    },
    {
      "epoch": 3.528066528066528,
      "grad_norm": 4.119503498077393,
      "learning_rate": 3.2732732732732736e-06,
      "loss": 0.1013,
      "step": 3394
    },
    {
      "epoch": 3.529106029106029,
      "grad_norm": 1.8827383518218994,
      "learning_rate": 3.270963270963271e-06,
      "loss": 0.0556,
      "step": 3395
    },
    {
      "epoch": 3.5301455301455302,
      "grad_norm": 1.0069574117660522,
      "learning_rate": 3.268653268653269e-06,
      "loss": 0.016,
      "step": 3396
    },
    {
      "epoch": 3.5311850311850312,
      "grad_norm": 2.0999436378479004,
      "learning_rate": 3.2663432663432665e-06,
      "loss": 0.0216,
      "step": 3397
    },
    {
      "epoch": 3.5322245322245323,
      "grad_norm": 11.844630241394043,
      "learning_rate": 3.2640332640332646e-06,
      "loss": 1.6668,
      "step": 3398
    },
    {
      "epoch": 3.5332640332640333,
      "grad_norm": 5.110901832580566,
      "learning_rate": 3.261723261723262e-06,
      "loss": 0.1361,
      "step": 3399
    },
    {
      "epoch": 3.5343035343035343,
      "grad_norm": 0.22296836972236633,
      "learning_rate": 3.2594132594132598e-06,
      "loss": 0.0056,
      "step": 3400
    },
    {
      "epoch": 3.5353430353430353,
      "grad_norm": 8.396090507507324,
      "learning_rate": 3.2571032571032575e-06,
      "loss": 0.3264,
      "step": 3401
    },
    {
      "epoch": 3.5363825363825363,
      "grad_norm": 0.7088819742202759,
      "learning_rate": 3.254793254793255e-06,
      "loss": 0.013,
      "step": 3402
    },
    {
      "epoch": 3.5374220374220373,
      "grad_norm": 1.2826210260391235,
      "learning_rate": 3.2524832524832526e-06,
      "loss": 0.0232,
      "step": 3403
    },
    {
      "epoch": 3.5384615384615383,
      "grad_norm": 8.47658634185791,
      "learning_rate": 3.2501732501732504e-06,
      "loss": 0.3357,
      "step": 3404
    },
    {
      "epoch": 3.5395010395010393,
      "grad_norm": 11.396049499511719,
      "learning_rate": 3.247863247863248e-06,
      "loss": 1.7146,
      "step": 3405
    },
    {
      "epoch": 3.5405405405405403,
      "grad_norm": 0.4763636291027069,
      "learning_rate": 3.2455532455532455e-06,
      "loss": 0.0211,
      "step": 3406
    },
    {
      "epoch": 3.5415800415800414,
      "grad_norm": 1.6495420932769775,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 0.0316,
      "step": 3407
    },
    {
      "epoch": 3.5426195426195424,
      "grad_norm": 2.4308693408966064,
      "learning_rate": 3.240933240933241e-06,
      "loss": 0.0561,
      "step": 3408
    },
    {
      "epoch": 3.5436590436590434,
      "grad_norm": 8.214385032653809,
      "learning_rate": 3.2386232386232392e-06,
      "loss": 0.3248,
      "step": 3409
    },
    {
      "epoch": 3.544698544698545,
      "grad_norm": 0.19386696815490723,
      "learning_rate": 3.2363132363132366e-06,
      "loss": 0.0022,
      "step": 3410
    },
    {
      "epoch": 3.545738045738046,
      "grad_norm": 6.6452107429504395,
      "learning_rate": 3.234003234003234e-06,
      "loss": 0.3384,
      "step": 3411
    },
    {
      "epoch": 3.546777546777547,
      "grad_norm": 7.752755165100098,
      "learning_rate": 3.231693231693232e-06,
      "loss": 0.7156,
      "step": 3412
    },
    {
      "epoch": 3.547817047817048,
      "grad_norm": 2.214608669281006,
      "learning_rate": 3.2293832293832295e-06,
      "loss": 0.0791,
      "step": 3413
    },
    {
      "epoch": 3.548856548856549,
      "grad_norm": 8.558451652526855,
      "learning_rate": 3.2270732270732276e-06,
      "loss": 0.9333,
      "step": 3414
    },
    {
      "epoch": 3.54989604989605,
      "grad_norm": 7.606333255767822,
      "learning_rate": 3.224763224763225e-06,
      "loss": 0.5964,
      "step": 3415
    },
    {
      "epoch": 3.550935550935551,
      "grad_norm": 3.464120388031006,
      "learning_rate": 3.2224532224532228e-06,
      "loss": 0.0738,
      "step": 3416
    },
    {
      "epoch": 3.551975051975052,
      "grad_norm": 12.014719009399414,
      "learning_rate": 3.2201432201432205e-06,
      "loss": 1.1626,
      "step": 3417
    },
    {
      "epoch": 3.553014553014553,
      "grad_norm": 16.28420639038086,
      "learning_rate": 3.2178332178332183e-06,
      "loss": 0.7914,
      "step": 3418
    },
    {
      "epoch": 3.554054054054054,
      "grad_norm": 0.35258999466896057,
      "learning_rate": 3.2155232155232156e-06,
      "loss": 0.0122,
      "step": 3419
    },
    {
      "epoch": 3.555093555093555,
      "grad_norm": 0.008318168111145496,
      "learning_rate": 3.2132132132132134e-06,
      "loss": 0.0002,
      "step": 3420
    },
    {
      "epoch": 3.556133056133056,
      "grad_norm": 0.10127261281013489,
      "learning_rate": 3.210903210903211e-06,
      "loss": 0.0022,
      "step": 3421
    },
    {
      "epoch": 3.5571725571725574,
      "grad_norm": 1.6699411869049072,
      "learning_rate": 3.2085932085932085e-06,
      "loss": 0.0236,
      "step": 3422
    },
    {
      "epoch": 3.5582120582120584,
      "grad_norm": 0.19638171792030334,
      "learning_rate": 3.2062832062832067e-06,
      "loss": 0.0037,
      "step": 3423
    },
    {
      "epoch": 3.5592515592515594,
      "grad_norm": 0.19892263412475586,
      "learning_rate": 3.203973203973204e-06,
      "loss": 0.0042,
      "step": 3424
    },
    {
      "epoch": 3.5602910602910605,
      "grad_norm": 6.505924224853516,
      "learning_rate": 3.2016632016632022e-06,
      "loss": 0.1503,
      "step": 3425
    },
    {
      "epoch": 3.5613305613305615,
      "grad_norm": 0.04345586895942688,
      "learning_rate": 3.1993531993531996e-06,
      "loss": 0.0011,
      "step": 3426
    },
    {
      "epoch": 3.5623700623700625,
      "grad_norm": 7.214348793029785,
      "learning_rate": 3.1970431970431974e-06,
      "loss": 0.4071,
      "step": 3427
    },
    {
      "epoch": 3.5634095634095635,
      "grad_norm": 0.15629810094833374,
      "learning_rate": 3.194733194733195e-06,
      "loss": 0.0053,
      "step": 3428
    },
    {
      "epoch": 3.5644490644490645,
      "grad_norm": 7.527303695678711,
      "learning_rate": 3.1924231924231925e-06,
      "loss": 0.3416,
      "step": 3429
    },
    {
      "epoch": 3.5654885654885655,
      "grad_norm": 9.97258472442627,
      "learning_rate": 3.1901131901131902e-06,
      "loss": 0.3462,
      "step": 3430
    },
    {
      "epoch": 3.5665280665280665,
      "grad_norm": 1.385948896408081,
      "learning_rate": 3.187803187803188e-06,
      "loss": 0.0396,
      "step": 3431
    },
    {
      "epoch": 3.5675675675675675,
      "grad_norm": 1.7325775623321533,
      "learning_rate": 3.1854931854931858e-06,
      "loss": 0.0559,
      "step": 3432
    },
    {
      "epoch": 3.5686070686070686,
      "grad_norm": 2.2397871017456055,
      "learning_rate": 3.183183183183183e-06,
      "loss": 0.0434,
      "step": 3433
    },
    {
      "epoch": 3.5696465696465696,
      "grad_norm": 0.028923479840159416,
      "learning_rate": 3.1808731808731813e-06,
      "loss": 0.0007,
      "step": 3434
    },
    {
      "epoch": 3.5706860706860706,
      "grad_norm": 1.0595364570617676,
      "learning_rate": 3.1785631785631786e-06,
      "loss": 0.0283,
      "step": 3435
    },
    {
      "epoch": 3.5717255717255716,
      "grad_norm": 2.357923746109009,
      "learning_rate": 3.176253176253177e-06,
      "loss": 0.0457,
      "step": 3436
    },
    {
      "epoch": 3.5727650727650726,
      "grad_norm": 11.933266639709473,
      "learning_rate": 3.173943173943174e-06,
      "loss": 1.2562,
      "step": 3437
    },
    {
      "epoch": 3.5738045738045736,
      "grad_norm": 8.789453506469727,
      "learning_rate": 3.1716331716331715e-06,
      "loss": 0.8383,
      "step": 3438
    },
    {
      "epoch": 3.5748440748440746,
      "grad_norm": 8.93902587890625,
      "learning_rate": 3.1693231693231697e-06,
      "loss": 0.6527,
      "step": 3439
    },
    {
      "epoch": 3.5758835758835756,
      "grad_norm": 0.027036476880311966,
      "learning_rate": 3.167013167013167e-06,
      "loss": 0.0005,
      "step": 3440
    },
    {
      "epoch": 3.5769230769230766,
      "grad_norm": 0.13876822590827942,
      "learning_rate": 3.1647031647031652e-06,
      "loss": 0.0027,
      "step": 3441
    },
    {
      "epoch": 3.577962577962578,
      "grad_norm": 8.124290466308594,
      "learning_rate": 3.1623931623931626e-06,
      "loss": 0.6514,
      "step": 3442
    },
    {
      "epoch": 3.579002079002079,
      "grad_norm": 0.695641279220581,
      "learning_rate": 3.1600831600831604e-06,
      "loss": 0.0171,
      "step": 3443
    },
    {
      "epoch": 3.58004158004158,
      "grad_norm": 9.36235523223877,
      "learning_rate": 3.157773157773158e-06,
      "loss": 0.4286,
      "step": 3444
    },
    {
      "epoch": 3.581081081081081,
      "grad_norm": 0.017492927610874176,
      "learning_rate": 3.155463155463156e-06,
      "loss": 0.0003,
      "step": 3445
    },
    {
      "epoch": 3.582120582120582,
      "grad_norm": 5.884636402130127,
      "learning_rate": 3.1531531531531532e-06,
      "loss": 0.6446,
      "step": 3446
    },
    {
      "epoch": 3.583160083160083,
      "grad_norm": 13.913434028625488,
      "learning_rate": 3.150843150843151e-06,
      "loss": 0.6823,
      "step": 3447
    },
    {
      "epoch": 3.584199584199584,
      "grad_norm": 8.719658851623535,
      "learning_rate": 3.1485331485331488e-06,
      "loss": 0.5322,
      "step": 3448
    },
    {
      "epoch": 3.585239085239085,
      "grad_norm": 10.48001480102539,
      "learning_rate": 3.146223146223146e-06,
      "loss": 1.0481,
      "step": 3449
    },
    {
      "epoch": 3.586278586278586,
      "grad_norm": 6.402904510498047,
      "learning_rate": 3.1439131439131443e-06,
      "loss": 0.447,
      "step": 3450
    },
    {
      "epoch": 3.587318087318087,
      "grad_norm": 9.76417064666748,
      "learning_rate": 3.1416031416031416e-06,
      "loss": 0.6952,
      "step": 3451
    },
    {
      "epoch": 3.5883575883575882,
      "grad_norm": 2.7027227878570557,
      "learning_rate": 3.13929313929314e-06,
      "loss": 0.0502,
      "step": 3452
    },
    {
      "epoch": 3.5893970893970892,
      "grad_norm": 0.3677961528301239,
      "learning_rate": 3.136983136983137e-06,
      "loss": 0.0098,
      "step": 3453
    },
    {
      "epoch": 3.5904365904365907,
      "grad_norm": 0.6174468398094177,
      "learning_rate": 3.1346731346731354e-06,
      "loss": 0.0145,
      "step": 3454
    },
    {
      "epoch": 3.5914760914760917,
      "grad_norm": 9.135771751403809,
      "learning_rate": 3.1323631323631327e-06,
      "loss": 0.2257,
      "step": 3455
    },
    {
      "epoch": 3.5925155925155927,
      "grad_norm": 0.7872626185417175,
      "learning_rate": 3.13005313005313e-06,
      "loss": 0.0139,
      "step": 3456
    },
    {
      "epoch": 3.5935550935550937,
      "grad_norm": 4.767543792724609,
      "learning_rate": 3.1277431277431283e-06,
      "loss": 0.1294,
      "step": 3457
    },
    {
      "epoch": 3.5945945945945947,
      "grad_norm": 9.243647575378418,
      "learning_rate": 3.1254331254331256e-06,
      "loss": 0.5878,
      "step": 3458
    },
    {
      "epoch": 3.5956340956340958,
      "grad_norm": 0.5455932021141052,
      "learning_rate": 3.1231231231231234e-06,
      "loss": 0.0145,
      "step": 3459
    },
    {
      "epoch": 3.5966735966735968,
      "grad_norm": 8.225960731506348,
      "learning_rate": 3.120813120813121e-06,
      "loss": 0.6251,
      "step": 3460
    },
    {
      "epoch": 3.5977130977130978,
      "grad_norm": 23.434959411621094,
      "learning_rate": 3.118503118503119e-06,
      "loss": 1.1641,
      "step": 3461
    },
    {
      "epoch": 3.598752598752599,
      "grad_norm": 5.3083720207214355,
      "learning_rate": 3.1161931161931162e-06,
      "loss": 0.1499,
      "step": 3462
    },
    {
      "epoch": 3.5997920997921,
      "grad_norm": 0.2932235300540924,
      "learning_rate": 3.1138831138831144e-06,
      "loss": 0.0054,
      "step": 3463
    },
    {
      "epoch": 3.600831600831601,
      "grad_norm": 0.014458829537034035,
      "learning_rate": 3.1115731115731118e-06,
      "loss": 0.0004,
      "step": 3464
    },
    {
      "epoch": 3.601871101871102,
      "grad_norm": 3.100857734680176,
      "learning_rate": 3.109263109263109e-06,
      "loss": 0.0848,
      "step": 3465
    },
    {
      "epoch": 3.602910602910603,
      "grad_norm": 3.241119384765625,
      "learning_rate": 3.1069531069531073e-06,
      "loss": 0.1285,
      "step": 3466
    },
    {
      "epoch": 3.603950103950104,
      "grad_norm": 1.9484933614730835,
      "learning_rate": 3.1046431046431047e-06,
      "loss": 0.0507,
      "step": 3467
    },
    {
      "epoch": 3.604989604989605,
      "grad_norm": 14.662372589111328,
      "learning_rate": 3.102333102333103e-06,
      "loss": 0.925,
      "step": 3468
    },
    {
      "epoch": 3.606029106029106,
      "grad_norm": 9.647832870483398,
      "learning_rate": 3.1000231000231e-06,
      "loss": 0.7678,
      "step": 3469
    },
    {
      "epoch": 3.607068607068607,
      "grad_norm": 2.403639316558838,
      "learning_rate": 3.097713097713098e-06,
      "loss": 0.1773,
      "step": 3470
    },
    {
      "epoch": 3.608108108108108,
      "grad_norm": 4.800213813781738,
      "learning_rate": 3.0954030954030957e-06,
      "loss": 0.1514,
      "step": 3471
    },
    {
      "epoch": 3.609147609147609,
      "grad_norm": 0.20564277470111847,
      "learning_rate": 3.0930930930930935e-06,
      "loss": 0.0037,
      "step": 3472
    },
    {
      "epoch": 3.61018711018711,
      "grad_norm": 0.821073055267334,
      "learning_rate": 3.090783090783091e-06,
      "loss": 0.0293,
      "step": 3473
    },
    {
      "epoch": 3.6112266112266114,
      "grad_norm": 1.197895884513855,
      "learning_rate": 3.0884730884730886e-06,
      "loss": 0.0228,
      "step": 3474
    },
    {
      "epoch": 3.6122661122661124,
      "grad_norm": 1.6000584363937378,
      "learning_rate": 3.0861630861630864e-06,
      "loss": 0.0328,
      "step": 3475
    },
    {
      "epoch": 3.6133056133056134,
      "grad_norm": 3.514110803604126,
      "learning_rate": 3.0838530838530837e-06,
      "loss": 0.0724,
      "step": 3476
    },
    {
      "epoch": 3.6143451143451144,
      "grad_norm": 11.626545906066895,
      "learning_rate": 3.081543081543082e-06,
      "loss": 0.9721,
      "step": 3477
    },
    {
      "epoch": 3.6153846153846154,
      "grad_norm": 7.133696556091309,
      "learning_rate": 3.0792330792330792e-06,
      "loss": 0.5373,
      "step": 3478
    },
    {
      "epoch": 3.6164241164241164,
      "grad_norm": 6.3263983726501465,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 0.2538,
      "step": 3479
    },
    {
      "epoch": 3.6174636174636174,
      "grad_norm": 5.5645599365234375,
      "learning_rate": 3.0746130746130748e-06,
      "loss": 0.4539,
      "step": 3480
    },
    {
      "epoch": 3.6185031185031185,
      "grad_norm": 1.2195281982421875,
      "learning_rate": 3.072303072303073e-06,
      "loss": 0.0161,
      "step": 3481
    },
    {
      "epoch": 3.6195426195426195,
      "grad_norm": 11.107911109924316,
      "learning_rate": 3.0699930699930703e-06,
      "loss": 0.2437,
      "step": 3482
    },
    {
      "epoch": 3.6205821205821205,
      "grad_norm": 4.7963056564331055,
      "learning_rate": 3.0676830676830677e-06,
      "loss": 0.1774,
      "step": 3483
    },
    {
      "epoch": 3.6216216216216215,
      "grad_norm": 3.392280101776123,
      "learning_rate": 3.065373065373066e-06,
      "loss": 0.0688,
      "step": 3484
    },
    {
      "epoch": 3.6226611226611225,
      "grad_norm": 3.227189540863037,
      "learning_rate": 3.063063063063063e-06,
      "loss": 0.0418,
      "step": 3485
    },
    {
      "epoch": 3.623700623700624,
      "grad_norm": 2.165048360824585,
      "learning_rate": 3.060753060753061e-06,
      "loss": 0.0669,
      "step": 3486
    },
    {
      "epoch": 3.624740124740125,
      "grad_norm": 6.590515613555908,
      "learning_rate": 3.0584430584430587e-06,
      "loss": 0.1336,
      "step": 3487
    },
    {
      "epoch": 3.625779625779626,
      "grad_norm": 0.3319731056690216,
      "learning_rate": 3.0561330561330565e-06,
      "loss": 0.0038,
      "step": 3488
    },
    {
      "epoch": 3.626819126819127,
      "grad_norm": 7.134136199951172,
      "learning_rate": 3.053823053823054e-06,
      "loss": 0.3527,
      "step": 3489
    },
    {
      "epoch": 3.627858627858628,
      "grad_norm": 0.6251439452171326,
      "learning_rate": 3.051513051513052e-06,
      "loss": 0.0108,
      "step": 3490
    },
    {
      "epoch": 3.628898128898129,
      "grad_norm": 0.04119293764233589,
      "learning_rate": 3.0492030492030494e-06,
      "loss": 0.0009,
      "step": 3491
    },
    {
      "epoch": 3.62993762993763,
      "grad_norm": 4.981802463531494,
      "learning_rate": 3.0468930468930467e-06,
      "loss": 0.1096,
      "step": 3492
    },
    {
      "epoch": 3.630977130977131,
      "grad_norm": 0.7034261226654053,
      "learning_rate": 3.044583044583045e-06,
      "loss": 0.0151,
      "step": 3493
    },
    {
      "epoch": 3.632016632016632,
      "grad_norm": 2.9406309127807617,
      "learning_rate": 3.0422730422730422e-06,
      "loss": 0.0738,
      "step": 3494
    },
    {
      "epoch": 3.633056133056133,
      "grad_norm": 0.5052170753479004,
      "learning_rate": 3.0399630399630404e-06,
      "loss": 0.0072,
      "step": 3495
    },
    {
      "epoch": 3.634095634095634,
      "grad_norm": 0.02603205479681492,
      "learning_rate": 3.0376530376530378e-06,
      "loss": 0.0008,
      "step": 3496
    },
    {
      "epoch": 3.635135135135135,
      "grad_norm": 10.114480972290039,
      "learning_rate": 3.0353430353430356e-06,
      "loss": 1.4564,
      "step": 3497
    },
    {
      "epoch": 3.636174636174636,
      "grad_norm": 2.4543769359588623,
      "learning_rate": 3.0330330330330333e-06,
      "loss": 0.049,
      "step": 3498
    },
    {
      "epoch": 3.637214137214137,
      "grad_norm": 0.6489257216453552,
      "learning_rate": 3.030723030723031e-06,
      "loss": 0.0133,
      "step": 3499
    },
    {
      "epoch": 3.638253638253638,
      "grad_norm": 0.2170177847146988,
      "learning_rate": 3.028413028413029e-06,
      "loss": 0.0026,
      "step": 3500
    },
    {
      "epoch": 3.639293139293139,
      "grad_norm": 9.83189868927002,
      "learning_rate": 3.026103026103026e-06,
      "loss": 0.4348,
      "step": 3501
    },
    {
      "epoch": 3.64033264033264,
      "grad_norm": 0.019325939938426018,
      "learning_rate": 3.023793023793024e-06,
      "loss": 0.0002,
      "step": 3502
    },
    {
      "epoch": 3.641372141372141,
      "grad_norm": 1.3234479427337646,
      "learning_rate": 3.0214830214830217e-06,
      "loss": 0.0501,
      "step": 3503
    },
    {
      "epoch": 3.642411642411642,
      "grad_norm": 0.08140242099761963,
      "learning_rate": 3.0191730191730195e-06,
      "loss": 0.0018,
      "step": 3504
    },
    {
      "epoch": 3.643451143451143,
      "grad_norm": 14.23491382598877,
      "learning_rate": 3.016863016863017e-06,
      "loss": 1.7548,
      "step": 3505
    },
    {
      "epoch": 3.6444906444906446,
      "grad_norm": 2.7035605907440186,
      "learning_rate": 3.014553014553015e-06,
      "loss": 0.2122,
      "step": 3506
    },
    {
      "epoch": 3.6455301455301456,
      "grad_norm": 5.186462879180908,
      "learning_rate": 3.0122430122430124e-06,
      "loss": 0.213,
      "step": 3507
    },
    {
      "epoch": 3.6465696465696467,
      "grad_norm": 0.010343966074287891,
      "learning_rate": 3.0099330099330106e-06,
      "loss": 0.0003,
      "step": 3508
    },
    {
      "epoch": 3.6476091476091477,
      "grad_norm": 5.633244514465332,
      "learning_rate": 3.007623007623008e-06,
      "loss": 0.2276,
      "step": 3509
    },
    {
      "epoch": 3.6486486486486487,
      "grad_norm": 10.875540733337402,
      "learning_rate": 3.0053130053130053e-06,
      "loss": 0.4257,
      "step": 3510
    },
    {
      "epoch": 3.6496881496881497,
      "grad_norm": 1.6572810411453247,
      "learning_rate": 3.0030030030030034e-06,
      "loss": 0.0349,
      "step": 3511
    },
    {
      "epoch": 3.6507276507276507,
      "grad_norm": 0.6229730844497681,
      "learning_rate": 3.0006930006930008e-06,
      "loss": 0.0178,
      "step": 3512
    },
    {
      "epoch": 3.6517671517671517,
      "grad_norm": 5.359668731689453,
      "learning_rate": 2.9983829983829986e-06,
      "loss": 0.1755,
      "step": 3513
    },
    {
      "epoch": 3.6528066528066527,
      "grad_norm": 4.23331880569458,
      "learning_rate": 2.9960729960729963e-06,
      "loss": 0.3545,
      "step": 3514
    },
    {
      "epoch": 3.6538461538461537,
      "grad_norm": 0.015081733465194702,
      "learning_rate": 2.993762993762994e-06,
      "loss": 0.0003,
      "step": 3515
    },
    {
      "epoch": 3.6548856548856548,
      "grad_norm": 1.0635371208190918,
      "learning_rate": 2.9914529914529914e-06,
      "loss": 0.025,
      "step": 3516
    },
    {
      "epoch": 3.6559251559251558,
      "grad_norm": 1.5098602771759033,
      "learning_rate": 2.9891429891429896e-06,
      "loss": 0.0284,
      "step": 3517
    },
    {
      "epoch": 3.6569646569646572,
      "grad_norm": 7.799502849578857,
      "learning_rate": 2.986832986832987e-06,
      "loss": 0.282,
      "step": 3518
    },
    {
      "epoch": 3.6580041580041582,
      "grad_norm": 5.349996566772461,
      "learning_rate": 2.9845229845229843e-06,
      "loss": 0.0988,
      "step": 3519
    },
    {
      "epoch": 3.6590436590436592,
      "grad_norm": 13.843528747558594,
      "learning_rate": 2.9822129822129825e-06,
      "loss": 1.0991,
      "step": 3520
    },
    {
      "epoch": 3.6600831600831603,
      "grad_norm": 8.505522727966309,
      "learning_rate": 2.97990297990298e-06,
      "loss": 0.2774,
      "step": 3521
    },
    {
      "epoch": 3.6611226611226613,
      "grad_norm": 7.561004161834717,
      "learning_rate": 2.977592977592978e-06,
      "loss": 0.4512,
      "step": 3522
    },
    {
      "epoch": 3.6621621621621623,
      "grad_norm": 1.7828177213668823,
      "learning_rate": 2.9752829752829754e-06,
      "loss": 0.2971,
      "step": 3523
    },
    {
      "epoch": 3.6632016632016633,
      "grad_norm": 3.8652234077453613,
      "learning_rate": 2.9729729729729736e-06,
      "loss": 0.0334,
      "step": 3524
    },
    {
      "epoch": 3.6642411642411643,
      "grad_norm": 3.4428791999816895,
      "learning_rate": 2.970662970662971e-06,
      "loss": 0.0855,
      "step": 3525
    },
    {
      "epoch": 3.6652806652806653,
      "grad_norm": 7.088212013244629,
      "learning_rate": 2.9683529683529687e-06,
      "loss": 0.5074,
      "step": 3526
    },
    {
      "epoch": 3.6663201663201663,
      "grad_norm": 0.06194361299276352,
      "learning_rate": 2.9660429660429664e-06,
      "loss": 0.0014,
      "step": 3527
    },
    {
      "epoch": 3.6673596673596673,
      "grad_norm": 0.017604293301701546,
      "learning_rate": 2.963732963732964e-06,
      "loss": 0.0004,
      "step": 3528
    },
    {
      "epoch": 3.6683991683991684,
      "grad_norm": 0.9275417923927307,
      "learning_rate": 2.9614229614229616e-06,
      "loss": 0.0126,
      "step": 3529
    },
    {
      "epoch": 3.6694386694386694,
      "grad_norm": 3.7533137798309326,
      "learning_rate": 2.9591129591129593e-06,
      "loss": 0.0467,
      "step": 3530
    },
    {
      "epoch": 3.6704781704781704,
      "grad_norm": 12.880677223205566,
      "learning_rate": 2.956802956802957e-06,
      "loss": 1.2227,
      "step": 3531
    },
    {
      "epoch": 3.6715176715176714,
      "grad_norm": 2.595752000808716,
      "learning_rate": 2.9544929544929544e-06,
      "loss": 0.0513,
      "step": 3532
    },
    {
      "epoch": 3.6725571725571724,
      "grad_norm": 2.3721625804901123,
      "learning_rate": 2.9521829521829526e-06,
      "loss": 0.0381,
      "step": 3533
    },
    {
      "epoch": 3.6735966735966734,
      "grad_norm": 3.710249423980713,
      "learning_rate": 2.94987294987295e-06,
      "loss": 0.3444,
      "step": 3534
    },
    {
      "epoch": 3.6746361746361744,
      "grad_norm": 2.203608274459839,
      "learning_rate": 2.947562947562948e-06,
      "loss": 0.0293,
      "step": 3535
    },
    {
      "epoch": 3.6756756756756754,
      "grad_norm": 4.4015374183654785,
      "learning_rate": 2.9452529452529455e-06,
      "loss": 0.2799,
      "step": 3536
    },
    {
      "epoch": 3.6767151767151764,
      "grad_norm": 3.468798875808716,
      "learning_rate": 2.942942942942943e-06,
      "loss": 0.3088,
      "step": 3537
    },
    {
      "epoch": 3.677754677754678,
      "grad_norm": 3.099963903427124,
      "learning_rate": 2.940632940632941e-06,
      "loss": 0.1014,
      "step": 3538
    },
    {
      "epoch": 3.678794178794179,
      "grad_norm": 0.2617235779762268,
      "learning_rate": 2.9383229383229384e-06,
      "loss": 0.0064,
      "step": 3539
    },
    {
      "epoch": 3.67983367983368,
      "grad_norm": 1.5205084085464478,
      "learning_rate": 2.936012936012936e-06,
      "loss": 0.2208,
      "step": 3540
    },
    {
      "epoch": 3.680873180873181,
      "grad_norm": 0.017373790964484215,
      "learning_rate": 2.933702933702934e-06,
      "loss": 0.0002,
      "step": 3541
    },
    {
      "epoch": 3.681912681912682,
      "grad_norm": NaN,
      "learning_rate": 2.9313929313929317e-06,
      "loss": 0.0641,
      "step": 3542
    },
    {
      "epoch": 3.682952182952183,
      "grad_norm": 0.20491603016853333,
      "learning_rate": 2.929082929082929e-06,
      "loss": 0.0043,
      "step": 3543
    },
    {
      "epoch": 3.683991683991684,
      "grad_norm": 0.0069188824854791164,
      "learning_rate": 2.9267729267729272e-06,
      "loss": 0.0002,
      "step": 3544
    },
    {
      "epoch": 3.685031185031185,
      "grad_norm": 4.759822845458984,
      "learning_rate": 2.9244629244629246e-06,
      "loss": 0.1093,
      "step": 3545
    },
    {
      "epoch": 3.686070686070686,
      "grad_norm": 3.06587553024292,
      "learning_rate": 2.9221529221529223e-06,
      "loss": 0.0937,
      "step": 3546
    },
    {
      "epoch": 3.687110187110187,
      "grad_norm": 9.3523588180542,
      "learning_rate": 2.91984291984292e-06,
      "loss": 0.472,
      "step": 3547
    },
    {
      "epoch": 3.688149688149688,
      "grad_norm": 0.035290349274873734,
      "learning_rate": 2.9175329175329174e-06,
      "loss": 0.0011,
      "step": 3548
    },
    {
      "epoch": 3.689189189189189,
      "grad_norm": 12.437941551208496,
      "learning_rate": 2.9152229152229156e-06,
      "loss": 0.4781,
      "step": 3549
    },
    {
      "epoch": 3.6902286902286905,
      "grad_norm": 14.929848670959473,
      "learning_rate": 2.912912912912913e-06,
      "loss": 1.4515,
      "step": 3550
    },
    {
      "epoch": 3.6912681912681915,
      "grad_norm": 8.693596839904785,
      "learning_rate": 2.910602910602911e-06,
      "loss": 0.6999,
      "step": 3551
    },
    {
      "epoch": 3.6923076923076925,
      "grad_norm": 1.6712979078292847,
      "learning_rate": 2.9082929082929085e-06,
      "loss": 0.2589,
      "step": 3552
    },
    {
      "epoch": 3.6933471933471935,
      "grad_norm": 3.1328914165496826,
      "learning_rate": 2.9059829059829063e-06,
      "loss": 0.1631,
      "step": 3553
    },
    {
      "epoch": 3.6943866943866945,
      "grad_norm": 4.731451511383057,
      "learning_rate": 2.903672903672904e-06,
      "loss": 0.1442,
      "step": 3554
    },
    {
      "epoch": 3.6954261954261955,
      "grad_norm": 4.9439520835876465,
      "learning_rate": 2.9013629013629014e-06,
      "loss": 0.3415,
      "step": 3555
    },
    {
      "epoch": 3.6964656964656966,
      "grad_norm": 9.034040451049805,
      "learning_rate": 2.899052899052899e-06,
      "loss": 0.5509,
      "step": 3556
    },
    {
      "epoch": 3.6975051975051976,
      "grad_norm": 0.06901997327804565,
      "learning_rate": 2.896742896742897e-06,
      "loss": 0.0023,
      "step": 3557
    },
    {
      "epoch": 3.6985446985446986,
      "grad_norm": 0.17667238414287567,
      "learning_rate": 2.8944328944328947e-06,
      "loss": 0.0032,
      "step": 3558
    },
    {
      "epoch": 3.6995841995841996,
      "grad_norm": 5.443127632141113,
      "learning_rate": 2.892122892122892e-06,
      "loss": 0.5829,
      "step": 3559
    },
    {
      "epoch": 3.7006237006237006,
      "grad_norm": 8.582060813903809,
      "learning_rate": 2.8898128898128902e-06,
      "loss": 0.3604,
      "step": 3560
    },
    {
      "epoch": 3.7016632016632016,
      "grad_norm": 2.5995655059814453,
      "learning_rate": 2.8875028875028876e-06,
      "loss": 0.0486,
      "step": 3561
    },
    {
      "epoch": 3.7027027027027026,
      "grad_norm": 0.021831559017300606,
      "learning_rate": 2.8851928851928858e-06,
      "loss": 0.0006,
      "step": 3562
    },
    {
      "epoch": 3.7037422037422036,
      "grad_norm": 3.39544677734375,
      "learning_rate": 2.882882882882883e-06,
      "loss": 0.0849,
      "step": 3563
    },
    {
      "epoch": 3.7047817047817047,
      "grad_norm": 0.7509887218475342,
      "learning_rate": 2.8805728805728804e-06,
      "loss": 0.0133,
      "step": 3564
    },
    {
      "epoch": 3.7058212058212057,
      "grad_norm": 12.253210067749023,
      "learning_rate": 2.8782628782628786e-06,
      "loss": 0.7792,
      "step": 3565
    },
    {
      "epoch": 3.7068607068607067,
      "grad_norm": 14.429997444152832,
      "learning_rate": 2.875952875952876e-06,
      "loss": 0.5612,
      "step": 3566
    },
    {
      "epoch": 3.7079002079002077,
      "grad_norm": 13.459504127502441,
      "learning_rate": 2.873642873642874e-06,
      "loss": 1.1039,
      "step": 3567
    },
    {
      "epoch": 3.7089397089397087,
      "grad_norm": 2.6319940090179443,
      "learning_rate": 2.8713328713328715e-06,
      "loss": 0.0435,
      "step": 3568
    },
    {
      "epoch": 3.7099792099792097,
      "grad_norm": 3.980374336242676,
      "learning_rate": 2.8690228690228693e-06,
      "loss": 0.0248,
      "step": 3569
    },
    {
      "epoch": 3.711018711018711,
      "grad_norm": 0.9240010976791382,
      "learning_rate": 2.866712866712867e-06,
      "loss": 0.0189,
      "step": 3570
    },
    {
      "epoch": 3.712058212058212,
      "grad_norm": 1.6734356880187988,
      "learning_rate": 2.864402864402865e-06,
      "loss": 0.0638,
      "step": 3571
    },
    {
      "epoch": 3.713097713097713,
      "grad_norm": 0.5525492429733276,
      "learning_rate": 2.862092862092862e-06,
      "loss": 0.0075,
      "step": 3572
    },
    {
      "epoch": 3.714137214137214,
      "grad_norm": 1.808472752571106,
      "learning_rate": 2.85978285978286e-06,
      "loss": 0.052,
      "step": 3573
    },
    {
      "epoch": 3.715176715176715,
      "grad_norm": 1.0177438259124756,
      "learning_rate": 2.8574728574728577e-06,
      "loss": 0.0199,
      "step": 3574
    },
    {
      "epoch": 3.7162162162162162,
      "grad_norm": 1.5117632150650024,
      "learning_rate": 2.855162855162855e-06,
      "loss": 0.0383,
      "step": 3575
    },
    {
      "epoch": 3.7172557172557172,
      "grad_norm": 0.37571460008621216,
      "learning_rate": 2.8528528528528532e-06,
      "loss": 0.0113,
      "step": 3576
    },
    {
      "epoch": 3.7182952182952183,
      "grad_norm": 0.6104511022567749,
      "learning_rate": 2.8505428505428506e-06,
      "loss": 0.0209,
      "step": 3577
    },
    {
      "epoch": 3.7193347193347193,
      "grad_norm": 0.32047128677368164,
      "learning_rate": 2.8482328482328488e-06,
      "loss": 0.0072,
      "step": 3578
    },
    {
      "epoch": 3.7203742203742203,
      "grad_norm": 11.813075065612793,
      "learning_rate": 2.845922845922846e-06,
      "loss": 0.4715,
      "step": 3579
    },
    {
      "epoch": 3.7214137214137213,
      "grad_norm": 0.27141356468200684,
      "learning_rate": 2.843612843612844e-06,
      "loss": 0.0039,
      "step": 3580
    },
    {
      "epoch": 3.7224532224532223,
      "grad_norm": 10.95467758178711,
      "learning_rate": 2.8413028413028416e-06,
      "loss": 0.6977,
      "step": 3581
    },
    {
      "epoch": 3.7234927234927238,
      "grad_norm": 0.06070644408464432,
      "learning_rate": 2.838992838992839e-06,
      "loss": 0.001,
      "step": 3582
    },
    {
      "epoch": 3.7245322245322248,
      "grad_norm": 0.3406676650047302,
      "learning_rate": 2.8366828366828368e-06,
      "loss": 0.0073,
      "step": 3583
    },
    {
      "epoch": 3.725571725571726,
      "grad_norm": 3.348909378051758,
      "learning_rate": 2.8343728343728345e-06,
      "loss": 0.063,
      "step": 3584
    },
    {
      "epoch": 3.726611226611227,
      "grad_norm": 1.7469083070755005,
      "learning_rate": 2.8320628320628323e-06,
      "loss": 0.0402,
      "step": 3585
    },
    {
      "epoch": 3.727650727650728,
      "grad_norm": 7.333977222442627,
      "learning_rate": 2.8297528297528296e-06,
      "loss": 0.1989,
      "step": 3586
    },
    {
      "epoch": 3.728690228690229,
      "grad_norm": 10.271039962768555,
      "learning_rate": 2.827442827442828e-06,
      "loss": 0.8281,
      "step": 3587
    },
    {
      "epoch": 3.72972972972973,
      "grad_norm": 0.02635607309639454,
      "learning_rate": 2.825132825132825e-06,
      "loss": 0.0005,
      "step": 3588
    },
    {
      "epoch": 3.730769230769231,
      "grad_norm": 8.348944664001465,
      "learning_rate": 2.8228228228228234e-06,
      "loss": 0.4226,
      "step": 3589
    },
    {
      "epoch": 3.731808731808732,
      "grad_norm": 12.303922653198242,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 1.1037,
      "step": 3590
    },
    {
      "epoch": 3.732848232848233,
      "grad_norm": 6.568941593170166,
      "learning_rate": 2.818202818202818e-06,
      "loss": 0.1263,
      "step": 3591
    },
    {
      "epoch": 3.733887733887734,
      "grad_norm": 11.41336441040039,
      "learning_rate": 2.8158928158928162e-06,
      "loss": 0.4161,
      "step": 3592
    },
    {
      "epoch": 3.734927234927235,
      "grad_norm": 8.585813522338867,
      "learning_rate": 2.8135828135828136e-06,
      "loss": 0.9174,
      "step": 3593
    },
    {
      "epoch": 3.735966735966736,
      "grad_norm": 8.427745819091797,
      "learning_rate": 2.8112728112728118e-06,
      "loss": 0.44,
      "step": 3594
    },
    {
      "epoch": 3.737006237006237,
      "grad_norm": 0.2848096787929535,
      "learning_rate": 2.808962808962809e-06,
      "loss": 0.0046,
      "step": 3595
    },
    {
      "epoch": 3.738045738045738,
      "grad_norm": 0.09448783099651337,
      "learning_rate": 2.806652806652807e-06,
      "loss": 0.0008,
      "step": 3596
    },
    {
      "epoch": 3.739085239085239,
      "grad_norm": 2.5192325115203857,
      "learning_rate": 2.8043428043428046e-06,
      "loss": 0.0636,
      "step": 3597
    },
    {
      "epoch": 3.74012474012474,
      "grad_norm": 1.4547226428985596,
      "learning_rate": 2.8020328020328024e-06,
      "loss": 0.035,
      "step": 3598
    },
    {
      "epoch": 3.741164241164241,
      "grad_norm": 8.2759370803833,
      "learning_rate": 2.7997227997227998e-06,
      "loss": 0.5974,
      "step": 3599
    },
    {
      "epoch": 3.742203742203742,
      "grad_norm": 0.031137343496084213,
      "learning_rate": 2.7974127974127975e-06,
      "loss": 0.0007,
      "step": 3600
    },
    {
      "epoch": 3.743243243243243,
      "grad_norm": 0.04575344920158386,
      "learning_rate": 2.7951027951027953e-06,
      "loss": 0.0014,
      "step": 3601
    },
    {
      "epoch": 3.7442827442827444,
      "grad_norm": 0.6326460242271423,
      "learning_rate": 2.7927927927927926e-06,
      "loss": 0.0223,
      "step": 3602
    },
    {
      "epoch": 3.7453222453222454,
      "grad_norm": 2.685260772705078,
      "learning_rate": 2.790482790482791e-06,
      "loss": 0.1627,
      "step": 3603
    },
    {
      "epoch": 3.7463617463617465,
      "grad_norm": 13.41769027709961,
      "learning_rate": 2.788172788172788e-06,
      "loss": 1.2246,
      "step": 3604
    },
    {
      "epoch": 3.7474012474012475,
      "grad_norm": 0.7796103358268738,
      "learning_rate": 2.7858627858627864e-06,
      "loss": 0.0158,
      "step": 3605
    },
    {
      "epoch": 3.7484407484407485,
      "grad_norm": 11.729232788085938,
      "learning_rate": 2.7835527835527837e-06,
      "loss": 1.0726,
      "step": 3606
    },
    {
      "epoch": 3.7494802494802495,
      "grad_norm": 3.69287371635437,
      "learning_rate": 2.781242781242782e-06,
      "loss": 0.0925,
      "step": 3607
    },
    {
      "epoch": 3.7505197505197505,
      "grad_norm": 0.11060631275177002,
      "learning_rate": 2.7789327789327792e-06,
      "loss": 0.0026,
      "step": 3608
    },
    {
      "epoch": 3.7515592515592515,
      "grad_norm": 8.287408828735352,
      "learning_rate": 2.776622776622777e-06,
      "loss": 0.2696,
      "step": 3609
    },
    {
      "epoch": 3.7525987525987525,
      "grad_norm": 0.1551632434129715,
      "learning_rate": 2.7743127743127748e-06,
      "loss": 0.0041,
      "step": 3610
    },
    {
      "epoch": 3.7536382536382535,
      "grad_norm": 3.1916396617889404,
      "learning_rate": 2.772002772002772e-06,
      "loss": 0.0784,
      "step": 3611
    },
    {
      "epoch": 3.7546777546777546,
      "grad_norm": 1.6691946983337402,
      "learning_rate": 2.76969276969277e-06,
      "loss": 0.0507,
      "step": 3612
    },
    {
      "epoch": 3.7557172557172556,
      "grad_norm": 4.951547145843506,
      "learning_rate": 2.7673827673827677e-06,
      "loss": 0.4273,
      "step": 3613
    },
    {
      "epoch": 3.756756756756757,
      "grad_norm": 6.517636299133301,
      "learning_rate": 2.7650727650727654e-06,
      "loss": 0.3348,
      "step": 3614
    },
    {
      "epoch": 3.757796257796258,
      "grad_norm": 4.4977006912231445,
      "learning_rate": 2.7627627627627628e-06,
      "loss": 0.0738,
      "step": 3615
    },
    {
      "epoch": 3.758835758835759,
      "grad_norm": 0.9126572012901306,
      "learning_rate": 2.760452760452761e-06,
      "loss": 0.0165,
      "step": 3616
    },
    {
      "epoch": 3.75987525987526,
      "grad_norm": 6.097523212432861,
      "learning_rate": 2.7581427581427583e-06,
      "loss": 0.1942,
      "step": 3617
    },
    {
      "epoch": 3.760914760914761,
      "grad_norm": 9.230613708496094,
      "learning_rate": 2.7558327558327565e-06,
      "loss": 0.4356,
      "step": 3618
    },
    {
      "epoch": 3.761954261954262,
      "grad_norm": 1.694145917892456,
      "learning_rate": 2.753522753522754e-06,
      "loss": 0.0581,
      "step": 3619
    },
    {
      "epoch": 3.762993762993763,
      "grad_norm": 5.806911468505859,
      "learning_rate": 2.751212751212751e-06,
      "loss": 0.195,
      "step": 3620
    },
    {
      "epoch": 3.764033264033264,
      "grad_norm": 10.217206954956055,
      "learning_rate": 2.7489027489027494e-06,
      "loss": 0.9645,
      "step": 3621
    },
    {
      "epoch": 3.765072765072765,
      "grad_norm": 2.835890531539917,
      "learning_rate": 2.7465927465927467e-06,
      "loss": 0.0674,
      "step": 3622
    },
    {
      "epoch": 3.766112266112266,
      "grad_norm": 0.19575683772563934,
      "learning_rate": 2.7442827442827445e-06,
      "loss": 0.0026,
      "step": 3623
    },
    {
      "epoch": 3.767151767151767,
      "grad_norm": 1.5427911281585693,
      "learning_rate": 2.7419727419727422e-06,
      "loss": 0.0391,
      "step": 3624
    },
    {
      "epoch": 3.768191268191268,
      "grad_norm": 9.355785369873047,
      "learning_rate": 2.73966273966274e-06,
      "loss": 0.6473,
      "step": 3625
    },
    {
      "epoch": 3.769230769230769,
      "grad_norm": 0.5041698217391968,
      "learning_rate": 2.7373527373527374e-06,
      "loss": 0.015,
      "step": 3626
    },
    {
      "epoch": 3.77027027027027,
      "grad_norm": 1.6092287302017212,
      "learning_rate": 2.7350427350427355e-06,
      "loss": 0.0286,
      "step": 3627
    },
    {
      "epoch": 3.771309771309771,
      "grad_norm": 1.0374270677566528,
      "learning_rate": 2.732732732732733e-06,
      "loss": 0.026,
      "step": 3628
    },
    {
      "epoch": 3.772349272349272,
      "grad_norm": 5.282305717468262,
      "learning_rate": 2.7304227304227302e-06,
      "loss": 0.1368,
      "step": 3629
    },
    {
      "epoch": 3.773388773388773,
      "grad_norm": 1.2266831398010254,
      "learning_rate": 2.7281127281127284e-06,
      "loss": 0.0191,
      "step": 3630
    },
    {
      "epoch": 3.774428274428274,
      "grad_norm": 0.8408501148223877,
      "learning_rate": 2.7258027258027258e-06,
      "loss": 0.0172,
      "step": 3631
    },
    {
      "epoch": 3.7754677754677752,
      "grad_norm": 0.9763440489768982,
      "learning_rate": 2.723492723492724e-06,
      "loss": 0.0166,
      "step": 3632
    },
    {
      "epoch": 3.7765072765072762,
      "grad_norm": 6.578399181365967,
      "learning_rate": 2.7211827211827213e-06,
      "loss": 0.0925,
      "step": 3633
    },
    {
      "epoch": 3.7775467775467777,
      "grad_norm": 0.2096863090991974,
      "learning_rate": 2.7188727188727195e-06,
      "loss": 0.0038,
      "step": 3634
    },
    {
      "epoch": 3.7785862785862787,
      "grad_norm": 1.804506540298462,
      "learning_rate": 2.716562716562717e-06,
      "loss": 0.0344,
      "step": 3635
    },
    {
      "epoch": 3.7796257796257797,
      "grad_norm": 0.49040916562080383,
      "learning_rate": 2.7142527142527146e-06,
      "loss": 0.0119,
      "step": 3636
    },
    {
      "epoch": 3.7806652806652807,
      "grad_norm": 11.101481437683105,
      "learning_rate": 2.7119427119427124e-06,
      "loss": 1.0053,
      "step": 3637
    },
    {
      "epoch": 3.7817047817047817,
      "grad_norm": 0.007577314507216215,
      "learning_rate": 2.7096327096327097e-06,
      "loss": 0.0001,
      "step": 3638
    },
    {
      "epoch": 3.7827442827442828,
      "grad_norm": 2.290045976638794,
      "learning_rate": 2.7073227073227075e-06,
      "loss": 0.0554,
      "step": 3639
    },
    {
      "epoch": 3.7837837837837838,
      "grad_norm": 0.050055865198373795,
      "learning_rate": 2.7050127050127052e-06,
      "loss": 0.0008,
      "step": 3640
    },
    {
      "epoch": 3.784823284823285,
      "grad_norm": 10.9364013671875,
      "learning_rate": 2.702702702702703e-06,
      "loss": 1.6398,
      "step": 3641
    },
    {
      "epoch": 3.785862785862786,
      "grad_norm": 0.030551081523299217,
      "learning_rate": 2.7003927003927004e-06,
      "loss": 0.0008,
      "step": 3642
    },
    {
      "epoch": 3.786902286902287,
      "grad_norm": 5.652614593505859,
      "learning_rate": 2.6980826980826985e-06,
      "loss": 0.0763,
      "step": 3643
    },
    {
      "epoch": 3.787941787941788,
      "grad_norm": 0.8022517561912537,
      "learning_rate": 2.695772695772696e-06,
      "loss": 0.0202,
      "step": 3644
    },
    {
      "epoch": 3.788981288981289,
      "grad_norm": 0.5187711715698242,
      "learning_rate": 2.693462693462694e-06,
      "loss": 0.0166,
      "step": 3645
    },
    {
      "epoch": 3.7900207900207903,
      "grad_norm": 0.14020437002182007,
      "learning_rate": 2.6911526911526914e-06,
      "loss": 0.004,
      "step": 3646
    },
    {
      "epoch": 3.7910602910602913,
      "grad_norm": 0.45986345410346985,
      "learning_rate": 2.6888426888426888e-06,
      "loss": 0.0068,
      "step": 3647
    },
    {
      "epoch": 3.7920997920997923,
      "grad_norm": 11.233826637268066,
      "learning_rate": 2.686532686532687e-06,
      "loss": 0.6995,
      "step": 3648
    },
    {
      "epoch": 3.7931392931392933,
      "grad_norm": 2.0428266525268555,
      "learning_rate": 2.6842226842226843e-06,
      "loss": 0.0425,
      "step": 3649
    },
    {
      "epoch": 3.7941787941787943,
      "grad_norm": 2.4397714138031006,
      "learning_rate": 2.681912681912682e-06,
      "loss": 0.0566,
      "step": 3650
    },
    {
      "epoch": 3.7952182952182953,
      "grad_norm": 13.182563781738281,
      "learning_rate": 2.67960267960268e-06,
      "loss": 0.7635,
      "step": 3651
    },
    {
      "epoch": 3.7962577962577964,
      "grad_norm": 0.6618658900260925,
      "learning_rate": 2.6772926772926776e-06,
      "loss": 0.0128,
      "step": 3652
    },
    {
      "epoch": 3.7972972972972974,
      "grad_norm": 1.9590808153152466,
      "learning_rate": 2.6749826749826754e-06,
      "loss": 0.0343,
      "step": 3653
    },
    {
      "epoch": 3.7983367983367984,
      "grad_norm": 6.853428363800049,
      "learning_rate": 2.672672672672673e-06,
      "loss": 0.1584,
      "step": 3654
    },
    {
      "epoch": 3.7993762993762994,
      "grad_norm": 4.4361724853515625,
      "learning_rate": 2.6703626703626705e-06,
      "loss": 0.0722,
      "step": 3655
    },
    {
      "epoch": 3.8004158004158004,
      "grad_norm": 4.45872163772583,
      "learning_rate": 2.6680526680526683e-06,
      "loss": 0.0835,
      "step": 3656
    },
    {
      "epoch": 3.8014553014553014,
      "grad_norm": 4.59832239151001,
      "learning_rate": 2.665742665742666e-06,
      "loss": 0.1583,
      "step": 3657
    },
    {
      "epoch": 3.8024948024948024,
      "grad_norm": 12.465092658996582,
      "learning_rate": 2.6634326634326634e-06,
      "loss": 1.0707,
      "step": 3658
    },
    {
      "epoch": 3.8035343035343034,
      "grad_norm": 11.42121410369873,
      "learning_rate": 2.6611226611226616e-06,
      "loss": 0.7943,
      "step": 3659
    },
    {
      "epoch": 3.8045738045738045,
      "grad_norm": 7.620397567749023,
      "learning_rate": 2.658812658812659e-06,
      "loss": 0.2346,
      "step": 3660
    },
    {
      "epoch": 3.8056133056133055,
      "grad_norm": 0.00030691453139297664,
      "learning_rate": 2.656502656502657e-06,
      "loss": 0.0,
      "step": 3661
    },
    {
      "epoch": 3.8066528066528065,
      "grad_norm": 2.865452766418457,
      "learning_rate": 2.6541926541926544e-06,
      "loss": 0.0559,
      "step": 3662
    },
    {
      "epoch": 3.8076923076923075,
      "grad_norm": 4.418657302856445,
      "learning_rate": 2.651882651882652e-06,
      "loss": 0.1019,
      "step": 3663
    },
    {
      "epoch": 3.8087318087318085,
      "grad_norm": 4.617337703704834,
      "learning_rate": 2.64957264957265e-06,
      "loss": 0.0996,
      "step": 3664
    },
    {
      "epoch": 3.8097713097713095,
      "grad_norm": 2.217773675918579,
      "learning_rate": 2.6472626472626473e-06,
      "loss": 0.0625,
      "step": 3665
    },
    {
      "epoch": 3.810810810810811,
      "grad_norm": 4.007612228393555,
      "learning_rate": 2.644952644952645e-06,
      "loss": 0.0961,
      "step": 3666
    },
    {
      "epoch": 3.811850311850312,
      "grad_norm": 8.661126136779785,
      "learning_rate": 2.642642642642643e-06,
      "loss": 0.5064,
      "step": 3667
    },
    {
      "epoch": 3.812889812889813,
      "grad_norm": 0.30983543395996094,
      "learning_rate": 2.6403326403326406e-06,
      "loss": 0.005,
      "step": 3668
    },
    {
      "epoch": 3.813929313929314,
      "grad_norm": 0.18076816201210022,
      "learning_rate": 2.638022638022638e-06,
      "loss": 0.0038,
      "step": 3669
    },
    {
      "epoch": 3.814968814968815,
      "grad_norm": 7.025464057922363,
      "learning_rate": 2.635712635712636e-06,
      "loss": 0.1497,
      "step": 3670
    },
    {
      "epoch": 3.816008316008316,
      "grad_norm": 12.076990127563477,
      "learning_rate": 2.6334026334026335e-06,
      "loss": 1.2221,
      "step": 3671
    },
    {
      "epoch": 3.817047817047817,
      "grad_norm": 6.618215084075928,
      "learning_rate": 2.6310926310926317e-06,
      "loss": 0.2126,
      "step": 3672
    },
    {
      "epoch": 3.818087318087318,
      "grad_norm": 1.5214687585830688,
      "learning_rate": 2.628782628782629e-06,
      "loss": 0.0272,
      "step": 3673
    },
    {
      "epoch": 3.819126819126819,
      "grad_norm": 7.2195281982421875,
      "learning_rate": 2.6264726264726264e-06,
      "loss": 0.4445,
      "step": 3674
    },
    {
      "epoch": 3.82016632016632,
      "grad_norm": 8.39342975616455,
      "learning_rate": 2.6241626241626246e-06,
      "loss": 0.2886,
      "step": 3675
    },
    {
      "epoch": 3.821205821205821,
      "grad_norm": 0.33063164353370667,
      "learning_rate": 2.621852621852622e-06,
      "loss": 0.0077,
      "step": 3676
    },
    {
      "epoch": 3.822245322245322,
      "grad_norm": 12.577540397644043,
      "learning_rate": 2.61954261954262e-06,
      "loss": 0.26,
      "step": 3677
    },
    {
      "epoch": 3.8232848232848236,
      "grad_norm": 6.113277912139893,
      "learning_rate": 2.6172326172326174e-06,
      "loss": 0.1863,
      "step": 3678
    },
    {
      "epoch": 3.8243243243243246,
      "grad_norm": 8.560226440429688,
      "learning_rate": 2.614922614922615e-06,
      "loss": 0.355,
      "step": 3679
    },
    {
      "epoch": 3.8253638253638256,
      "grad_norm": 9.159148216247559,
      "learning_rate": 2.612612612612613e-06,
      "loss": 0.2278,
      "step": 3680
    },
    {
      "epoch": 3.8264033264033266,
      "grad_norm": 4.1118574142456055,
      "learning_rate": 2.6103026103026107e-06,
      "loss": 0.1376,
      "step": 3681
    },
    {
      "epoch": 3.8274428274428276,
      "grad_norm": 0.10904556512832642,
      "learning_rate": 2.607992607992608e-06,
      "loss": 0.0023,
      "step": 3682
    },
    {
      "epoch": 3.8284823284823286,
      "grad_norm": 0.13135097920894623,
      "learning_rate": 2.605682605682606e-06,
      "loss": 0.002,
      "step": 3683
    },
    {
      "epoch": 3.8295218295218296,
      "grad_norm": 13.160359382629395,
      "learning_rate": 2.6033726033726036e-06,
      "loss": 2.0968,
      "step": 3684
    },
    {
      "epoch": 3.8305613305613306,
      "grad_norm": 4.60862922668457,
      "learning_rate": 2.601062601062601e-06,
      "loss": 0.156,
      "step": 3685
    },
    {
      "epoch": 3.8316008316008316,
      "grad_norm": 1.0691585540771484,
      "learning_rate": 2.598752598752599e-06,
      "loss": 0.0158,
      "step": 3686
    },
    {
      "epoch": 3.8326403326403327,
      "grad_norm": 0.3119029700756073,
      "learning_rate": 2.5964425964425965e-06,
      "loss": 0.0057,
      "step": 3687
    },
    {
      "epoch": 3.8336798336798337,
      "grad_norm": 0.9196264743804932,
      "learning_rate": 2.5941325941325947e-06,
      "loss": 0.0313,
      "step": 3688
    },
    {
      "epoch": 3.8347193347193347,
      "grad_norm": 3.6135544776916504,
      "learning_rate": 2.591822591822592e-06,
      "loss": 0.0938,
      "step": 3689
    },
    {
      "epoch": 3.8357588357588357,
      "grad_norm": 5.493073463439941,
      "learning_rate": 2.58951258951259e-06,
      "loss": 0.2199,
      "step": 3690
    },
    {
      "epoch": 3.8367983367983367,
      "grad_norm": 0.0029417388141155243,
      "learning_rate": 2.5872025872025876e-06,
      "loss": 0.0001,
      "step": 3691
    },
    {
      "epoch": 3.8378378378378377,
      "grad_norm": 0.00937197357416153,
      "learning_rate": 2.584892584892585e-06,
      "loss": 0.0002,
      "step": 3692
    },
    {
      "epoch": 3.8388773388773387,
      "grad_norm": 7.578929424285889,
      "learning_rate": 2.5825825825825827e-06,
      "loss": 0.3299,
      "step": 3693
    },
    {
      "epoch": 3.8399168399168397,
      "grad_norm": 9.87551212310791,
      "learning_rate": 2.5802725802725804e-06,
      "loss": 0.6378,
      "step": 3694
    },
    {
      "epoch": 3.8409563409563408,
      "grad_norm": 0.6146271824836731,
      "learning_rate": 2.577962577962578e-06,
      "loss": 0.0142,
      "step": 3695
    },
    {
      "epoch": 3.8419958419958418,
      "grad_norm": 1.575995683670044,
      "learning_rate": 2.5756525756525756e-06,
      "loss": 0.0283,
      "step": 3696
    },
    {
      "epoch": 3.8430353430353428,
      "grad_norm": 0.2135927826166153,
      "learning_rate": 2.5733425733425737e-06,
      "loss": 0.0048,
      "step": 3697
    },
    {
      "epoch": 3.8440748440748442,
      "grad_norm": 0.813083827495575,
      "learning_rate": 2.571032571032571e-06,
      "loss": 0.0203,
      "step": 3698
    },
    {
      "epoch": 3.8451143451143452,
      "grad_norm": 6.919641971588135,
      "learning_rate": 2.5687225687225693e-06,
      "loss": 1.2358,
      "step": 3699
    },
    {
      "epoch": 3.8461538461538463,
      "grad_norm": 7.1949028968811035,
      "learning_rate": 2.5664125664125666e-06,
      "loss": 0.1928,
      "step": 3700
    },
    {
      "epoch": 3.8471933471933473,
      "grad_norm": 10.908730506896973,
      "learning_rate": 2.564102564102564e-06,
      "loss": 1.4967,
      "step": 3701
    },
    {
      "epoch": 3.8482328482328483,
      "grad_norm": 1.6860178709030151,
      "learning_rate": 2.561792561792562e-06,
      "loss": 0.1061,
      "step": 3702
    },
    {
      "epoch": 3.8492723492723493,
      "grad_norm": 1.1886787414550781,
      "learning_rate": 2.5594825594825595e-06,
      "loss": 0.0334,
      "step": 3703
    },
    {
      "epoch": 3.8503118503118503,
      "grad_norm": 0.08556528389453888,
      "learning_rate": 2.5571725571725577e-06,
      "loss": 0.0013,
      "step": 3704
    },
    {
      "epoch": 3.8513513513513513,
      "grad_norm": 1.3318017721176147,
      "learning_rate": 2.554862554862555e-06,
      "loss": 0.0475,
      "step": 3705
    },
    {
      "epoch": 3.8523908523908523,
      "grad_norm": 0.05544210225343704,
      "learning_rate": 2.552552552552553e-06,
      "loss": 0.0007,
      "step": 3706
    },
    {
      "epoch": 3.8534303534303533,
      "grad_norm": 0.11199303716421127,
      "learning_rate": 2.5502425502425506e-06,
      "loss": 0.0022,
      "step": 3707
    },
    {
      "epoch": 3.8544698544698544,
      "grad_norm": 4.047089099884033,
      "learning_rate": 2.5479325479325483e-06,
      "loss": 0.0905,
      "step": 3708
    },
    {
      "epoch": 3.8555093555093554,
      "grad_norm": 4.6328935623168945,
      "learning_rate": 2.5456225456225457e-06,
      "loss": 0.074,
      "step": 3709
    },
    {
      "epoch": 3.856548856548857,
      "grad_norm": 15.714149475097656,
      "learning_rate": 2.5433125433125434e-06,
      "loss": 0.0865,
      "step": 3710
    },
    {
      "epoch": 3.857588357588358,
      "grad_norm": 0.0857136994600296,
      "learning_rate": 2.5410025410025412e-06,
      "loss": 0.0016,
      "step": 3711
    },
    {
      "epoch": 3.858627858627859,
      "grad_norm": 3.9635956287384033,
      "learning_rate": 2.5386925386925386e-06,
      "loss": 0.2832,
      "step": 3712
    },
    {
      "epoch": 3.85966735966736,
      "grad_norm": 2.428004741668701,
      "learning_rate": 2.5363825363825367e-06,
      "loss": 0.0717,
      "step": 3713
    },
    {
      "epoch": 3.860706860706861,
      "grad_norm": 0.019456250593066216,
      "learning_rate": 2.534072534072534e-06,
      "loss": 0.0002,
      "step": 3714
    },
    {
      "epoch": 3.861746361746362,
      "grad_norm": 0.16324344277381897,
      "learning_rate": 2.5317625317625323e-06,
      "loss": 0.0036,
      "step": 3715
    },
    {
      "epoch": 3.862785862785863,
      "grad_norm": 8.783102035522461,
      "learning_rate": 2.5294525294525296e-06,
      "loss": 0.4462,
      "step": 3716
    },
    {
      "epoch": 3.863825363825364,
      "grad_norm": 1.616220474243164,
      "learning_rate": 2.527142527142528e-06,
      "loss": 0.0132,
      "step": 3717
    },
    {
      "epoch": 3.864864864864865,
      "grad_norm": 0.14159680902957916,
      "learning_rate": 2.524832524832525e-06,
      "loss": 0.0042,
      "step": 3718
    },
    {
      "epoch": 3.865904365904366,
      "grad_norm": 1.1944466829299927,
      "learning_rate": 2.5225225225225225e-06,
      "loss": 0.0295,
      "step": 3719
    },
    {
      "epoch": 3.866943866943867,
      "grad_norm": 1.1400628089904785,
      "learning_rate": 2.5202125202125207e-06,
      "loss": 0.0242,
      "step": 3720
    },
    {
      "epoch": 3.867983367983368,
      "grad_norm": 0.20073829591274261,
      "learning_rate": 2.517902517902518e-06,
      "loss": 0.0039,
      "step": 3721
    },
    {
      "epoch": 3.869022869022869,
      "grad_norm": 5.438026428222656,
      "learning_rate": 2.515592515592516e-06,
      "loss": 0.1886,
      "step": 3722
    },
    {
      "epoch": 3.87006237006237,
      "grad_norm": 0.04271695762872696,
      "learning_rate": 2.5132825132825136e-06,
      "loss": 0.0006,
      "step": 3723
    },
    {
      "epoch": 3.871101871101871,
      "grad_norm": 4.7715654373168945,
      "learning_rate": 2.5109725109725113e-06,
      "loss": 0.3594,
      "step": 3724
    },
    {
      "epoch": 3.872141372141372,
      "grad_norm": 0.6592026352882385,
      "learning_rate": 2.5086625086625087e-06,
      "loss": 0.0126,
      "step": 3725
    },
    {
      "epoch": 3.873180873180873,
      "grad_norm": 0.3664993941783905,
      "learning_rate": 2.506352506352507e-06,
      "loss": 0.0105,
      "step": 3726
    },
    {
      "epoch": 3.874220374220374,
      "grad_norm": 7.50433349609375,
      "learning_rate": 2.5040425040425042e-06,
      "loss": 0.2574,
      "step": 3727
    },
    {
      "epoch": 3.875259875259875,
      "grad_norm": 9.310280799865723,
      "learning_rate": 2.5017325017325016e-06,
      "loss": 0.286,
      "step": 3728
    },
    {
      "epoch": 3.876299376299376,
      "grad_norm": 7.536202907562256,
      "learning_rate": 2.4994224994224998e-06,
      "loss": 0.3656,
      "step": 3729
    },
    {
      "epoch": 3.8773388773388775,
      "grad_norm": 0.292306512594223,
      "learning_rate": 2.4971124971124975e-06,
      "loss": 0.0058,
      "step": 3730
    },
    {
      "epoch": 3.8783783783783785,
      "grad_norm": 1.817397952079773,
      "learning_rate": 2.4948024948024953e-06,
      "loss": 0.0895,
      "step": 3731
    },
    {
      "epoch": 3.8794178794178795,
      "grad_norm": 3.750711679458618,
      "learning_rate": 2.4924924924924926e-06,
      "loss": 0.0859,
      "step": 3732
    },
    {
      "epoch": 3.8804573804573805,
      "grad_norm": 9.821932792663574,
      "learning_rate": 2.4901824901824904e-06,
      "loss": 0.7896,
      "step": 3733
    },
    {
      "epoch": 3.8814968814968815,
      "grad_norm": 0.03397275507450104,
      "learning_rate": 2.487872487872488e-06,
      "loss": 0.0007,
      "step": 3734
    },
    {
      "epoch": 3.8825363825363826,
      "grad_norm": 7.19288969039917,
      "learning_rate": 2.485562485562486e-06,
      "loss": 0.468,
      "step": 3735
    },
    {
      "epoch": 3.8835758835758836,
      "grad_norm": 9.314663887023926,
      "learning_rate": 2.4832524832524833e-06,
      "loss": 0.7014,
      "step": 3736
    },
    {
      "epoch": 3.8846153846153846,
      "grad_norm": 3.0756399631500244,
      "learning_rate": 2.480942480942481e-06,
      "loss": 0.0745,
      "step": 3737
    },
    {
      "epoch": 3.8856548856548856,
      "grad_norm": 2.730370283126831,
      "learning_rate": 2.478632478632479e-06,
      "loss": 0.0803,
      "step": 3738
    },
    {
      "epoch": 3.8866943866943866,
      "grad_norm": 5.826650142669678,
      "learning_rate": 2.4763224763224766e-06,
      "loss": 0.6514,
      "step": 3739
    },
    {
      "epoch": 3.8877338877338876,
      "grad_norm": 2.1858022212982178,
      "learning_rate": 2.4740124740124743e-06,
      "loss": 0.0215,
      "step": 3740
    },
    {
      "epoch": 3.8887733887733886,
      "grad_norm": 3.4572198390960693,
      "learning_rate": 2.4717024717024717e-06,
      "loss": 0.0368,
      "step": 3741
    },
    {
      "epoch": 3.88981288981289,
      "grad_norm": 0.550865650177002,
      "learning_rate": 2.4693924693924695e-06,
      "loss": 0.0119,
      "step": 3742
    },
    {
      "epoch": 3.890852390852391,
      "grad_norm": 2.629605293273926,
      "learning_rate": 2.4670824670824672e-06,
      "loss": 0.0367,
      "step": 3743
    },
    {
      "epoch": 3.891891891891892,
      "grad_norm": 12.43010139465332,
      "learning_rate": 2.464772464772465e-06,
      "loss": 1.614,
      "step": 3744
    },
    {
      "epoch": 3.892931392931393,
      "grad_norm": 0.385886013507843,
      "learning_rate": 2.4624624624624628e-06,
      "loss": 0.0111,
      "step": 3745
    },
    {
      "epoch": 3.893970893970894,
      "grad_norm": 0.2466421276330948,
      "learning_rate": 2.4601524601524605e-06,
      "loss": 0.0066,
      "step": 3746
    },
    {
      "epoch": 3.895010395010395,
      "grad_norm": 0.9171250462532043,
      "learning_rate": 2.4578424578424583e-06,
      "loss": 0.2473,
      "step": 3747
    },
    {
      "epoch": 3.896049896049896,
      "grad_norm": 0.02100229635834694,
      "learning_rate": 2.4555324555324556e-06,
      "loss": 0.0006,
      "step": 3748
    },
    {
      "epoch": 3.897089397089397,
      "grad_norm": 0.4232587218284607,
      "learning_rate": 2.4532224532224534e-06,
      "loss": 0.0099,
      "step": 3749
    },
    {
      "epoch": 3.898128898128898,
      "grad_norm": 0.06537114083766937,
      "learning_rate": 2.450912450912451e-06,
      "loss": 0.0018,
      "step": 3750
    },
    {
      "epoch": 3.899168399168399,
      "grad_norm": 0.0012716350611299276,
      "learning_rate": 2.4486024486024485e-06,
      "loss": 0.0,
      "step": 3751
    },
    {
      "epoch": 3.9002079002079,
      "grad_norm": 8.932744026184082,
      "learning_rate": 2.4462924462924463e-06,
      "loss": 1.0465,
      "step": 3752
    },
    {
      "epoch": 3.901247401247401,
      "grad_norm": 3.2070016860961914,
      "learning_rate": 2.443982443982444e-06,
      "loss": 0.0669,
      "step": 3753
    },
    {
      "epoch": 3.9022869022869022,
      "grad_norm": 0.04936572164297104,
      "learning_rate": 2.441672441672442e-06,
      "loss": 0.0012,
      "step": 3754
    },
    {
      "epoch": 3.9033264033264032,
      "grad_norm": 0.60624098777771,
      "learning_rate": 2.4393624393624396e-06,
      "loss": 0.0073,
      "step": 3755
    },
    {
      "epoch": 3.9043659043659042,
      "grad_norm": 3.0943517684936523,
      "learning_rate": 2.4370524370524373e-06,
      "loss": 0.0836,
      "step": 3756
    },
    {
      "epoch": 3.9054054054054053,
      "grad_norm": 0.1718333512544632,
      "learning_rate": 2.434742434742435e-06,
      "loss": 0.0023,
      "step": 3757
    },
    {
      "epoch": 3.9064449064449063,
      "grad_norm": 0.06529901921749115,
      "learning_rate": 2.432432432432433e-06,
      "loss": 0.0009,
      "step": 3758
    },
    {
      "epoch": 3.9074844074844073,
      "grad_norm": 11.549318313598633,
      "learning_rate": 2.4301224301224302e-06,
      "loss": 0.5639,
      "step": 3759
    },
    {
      "epoch": 3.9085239085239083,
      "grad_norm": 0.7153767347335815,
      "learning_rate": 2.427812427812428e-06,
      "loss": 0.0206,
      "step": 3760
    },
    {
      "epoch": 3.9095634095634093,
      "grad_norm": 1.1701185703277588,
      "learning_rate": 2.4255024255024258e-06,
      "loss": 0.0226,
      "step": 3761
    },
    {
      "epoch": 3.9106029106029108,
      "grad_norm": 1.5243420600891113,
      "learning_rate": 2.4231924231924235e-06,
      "loss": 0.0257,
      "step": 3762
    },
    {
      "epoch": 3.9116424116424118,
      "grad_norm": 9.948197364807129,
      "learning_rate": 2.4208824208824213e-06,
      "loss": 0.527,
      "step": 3763
    },
    {
      "epoch": 3.912681912681913,
      "grad_norm": 0.8991334438323975,
      "learning_rate": 2.4185724185724186e-06,
      "loss": 0.0239,
      "step": 3764
    },
    {
      "epoch": 3.913721413721414,
      "grad_norm": 22.073930740356445,
      "learning_rate": 2.4162624162624164e-06,
      "loss": 1.0042,
      "step": 3765
    },
    {
      "epoch": 3.914760914760915,
      "grad_norm": 0.10600987076759338,
      "learning_rate": 2.413952413952414e-06,
      "loss": 0.0013,
      "step": 3766
    },
    {
      "epoch": 3.915800415800416,
      "grad_norm": 0.7799251675605774,
      "learning_rate": 2.411642411642412e-06,
      "loss": 0.0164,
      "step": 3767
    },
    {
      "epoch": 3.916839916839917,
      "grad_norm": 8.539443969726562,
      "learning_rate": 2.4093324093324093e-06,
      "loss": 0.2256,
      "step": 3768
    },
    {
      "epoch": 3.917879417879418,
      "grad_norm": 8.018176078796387,
      "learning_rate": 2.407022407022407e-06,
      "loss": 0.567,
      "step": 3769
    },
    {
      "epoch": 3.918918918918919,
      "grad_norm": 2.124363422393799,
      "learning_rate": 2.404712404712405e-06,
      "loss": 0.3356,
      "step": 3770
    },
    {
      "epoch": 3.91995841995842,
      "grad_norm": 7.096327304840088,
      "learning_rate": 2.4024024024024026e-06,
      "loss": 0.3187,
      "step": 3771
    },
    {
      "epoch": 3.920997920997921,
      "grad_norm": 8.52831745147705,
      "learning_rate": 2.4000924000924004e-06,
      "loss": 0.6401,
      "step": 3772
    },
    {
      "epoch": 3.922037422037422,
      "grad_norm": 2.970491886138916,
      "learning_rate": 2.397782397782398e-06,
      "loss": 0.0845,
      "step": 3773
    },
    {
      "epoch": 3.9230769230769234,
      "grad_norm": 0.8159802556037903,
      "learning_rate": 2.395472395472396e-06,
      "loss": 0.0412,
      "step": 3774
    },
    {
      "epoch": 3.9241164241164244,
      "grad_norm": 0.5145953893661499,
      "learning_rate": 2.3931623931623937e-06,
      "loss": 0.0138,
      "step": 3775
    },
    {
      "epoch": 3.9251559251559254,
      "grad_norm": 0.3560482859611511,
      "learning_rate": 2.390852390852391e-06,
      "loss": 0.0075,
      "step": 3776
    },
    {
      "epoch": 3.9261954261954264,
      "grad_norm": 14.716246604919434,
      "learning_rate": 2.3885423885423888e-06,
      "loss": 1.043,
      "step": 3777
    },
    {
      "epoch": 3.9272349272349274,
      "grad_norm": 0.0786583423614502,
      "learning_rate": 2.3862323862323865e-06,
      "loss": 0.0011,
      "step": 3778
    },
    {
      "epoch": 3.9282744282744284,
      "grad_norm": 14.384025573730469,
      "learning_rate": 2.383922383922384e-06,
      "loss": 0.8149,
      "step": 3779
    },
    {
      "epoch": 3.9293139293139294,
      "grad_norm": 1.1967740058898926,
      "learning_rate": 2.3816123816123816e-06,
      "loss": 0.0347,
      "step": 3780
    },
    {
      "epoch": 3.9303534303534304,
      "grad_norm": 6.0211381912231445,
      "learning_rate": 2.3793023793023794e-06,
      "loss": 0.1471,
      "step": 3781
    },
    {
      "epoch": 3.9313929313929314,
      "grad_norm": 6.998178482055664,
      "learning_rate": 2.376992376992377e-06,
      "loss": 0.2397,
      "step": 3782
    },
    {
      "epoch": 3.9324324324324325,
      "grad_norm": 0.5392391085624695,
      "learning_rate": 2.374682374682375e-06,
      "loss": 0.011,
      "step": 3783
    },
    {
      "epoch": 3.9334719334719335,
      "grad_norm": 0.8353741765022278,
      "learning_rate": 2.3723723723723727e-06,
      "loss": 0.0045,
      "step": 3784
    },
    {
      "epoch": 3.9345114345114345,
      "grad_norm": 0.29350486397743225,
      "learning_rate": 2.3700623700623705e-06,
      "loss": 0.005,
      "step": 3785
    },
    {
      "epoch": 3.9355509355509355,
      "grad_norm": 0.05252784863114357,
      "learning_rate": 2.367752367752368e-06,
      "loss": 0.0017,
      "step": 3786
    },
    {
      "epoch": 3.9365904365904365,
      "grad_norm": 0.7203182578086853,
      "learning_rate": 2.3654423654423656e-06,
      "loss": 0.0167,
      "step": 3787
    },
    {
      "epoch": 3.9376299376299375,
      "grad_norm": 0.04804070293903351,
      "learning_rate": 2.3631323631323634e-06,
      "loss": 0.0011,
      "step": 3788
    },
    {
      "epoch": 3.9386694386694385,
      "grad_norm": 7.088498592376709,
      "learning_rate": 2.360822360822361e-06,
      "loss": 0.6737,
      "step": 3789
    },
    {
      "epoch": 3.9397089397089395,
      "grad_norm": 0.023920787498354912,
      "learning_rate": 2.358512358512359e-06,
      "loss": 0.0006,
      "step": 3790
    },
    {
      "epoch": 3.9407484407484406,
      "grad_norm": 15.3619966506958,
      "learning_rate": 2.3562023562023562e-06,
      "loss": 0.9882,
      "step": 3791
    },
    {
      "epoch": 3.9417879417879416,
      "grad_norm": 1.5771281719207764,
      "learning_rate": 2.353892353892354e-06,
      "loss": 0.0176,
      "step": 3792
    },
    {
      "epoch": 3.9428274428274426,
      "grad_norm": 11.848381042480469,
      "learning_rate": 2.3515823515823518e-06,
      "loss": 1.0265,
      "step": 3793
    },
    {
      "epoch": 3.943866943866944,
      "grad_norm": 11.604540824890137,
      "learning_rate": 2.3492723492723495e-06,
      "loss": 1.1948,
      "step": 3794
    },
    {
      "epoch": 3.944906444906445,
      "grad_norm": 12.787267684936523,
      "learning_rate": 2.346962346962347e-06,
      "loss": 1.0989,
      "step": 3795
    },
    {
      "epoch": 3.945945945945946,
      "grad_norm": 5.049587249755859,
      "learning_rate": 2.3446523446523446e-06,
      "loss": 0.2074,
      "step": 3796
    },
    {
      "epoch": 3.946985446985447,
      "grad_norm": 0.019799979403614998,
      "learning_rate": 2.3423423423423424e-06,
      "loss": 0.0004,
      "step": 3797
    },
    {
      "epoch": 3.948024948024948,
      "grad_norm": 0.018526798114180565,
      "learning_rate": 2.34003234003234e-06,
      "loss": 0.0003,
      "step": 3798
    },
    {
      "epoch": 3.949064449064449,
      "grad_norm": 0.3677530884742737,
      "learning_rate": 2.337722337722338e-06,
      "loss": 0.0139,
      "step": 3799
    },
    {
      "epoch": 3.95010395010395,
      "grad_norm": 0.17251178622245789,
      "learning_rate": 2.3354123354123357e-06,
      "loss": 0.0039,
      "step": 3800
    },
    {
      "epoch": 3.951143451143451,
      "grad_norm": 9.902690887451172,
      "learning_rate": 2.3331023331023335e-06,
      "loss": 0.5293,
      "step": 3801
    },
    {
      "epoch": 3.952182952182952,
      "grad_norm": 0.6778705716133118,
      "learning_rate": 2.3307923307923313e-06,
      "loss": 0.0143,
      "step": 3802
    },
    {
      "epoch": 3.953222453222453,
      "grad_norm": 0.0005741470959037542,
      "learning_rate": 2.3284823284823286e-06,
      "loss": 0.0,
      "step": 3803
    },
    {
      "epoch": 3.954261954261954,
      "grad_norm": 15.404338836669922,
      "learning_rate": 2.3261723261723264e-06,
      "loss": 0.6453,
      "step": 3804
    },
    {
      "epoch": 3.955301455301455,
      "grad_norm": 7.858543395996094,
      "learning_rate": 2.323862323862324e-06,
      "loss": 0.296,
      "step": 3805
    },
    {
      "epoch": 3.9563409563409566,
      "grad_norm": 4.857236385345459,
      "learning_rate": 2.321552321552322e-06,
      "loss": 0.1043,
      "step": 3806
    },
    {
      "epoch": 3.9573804573804576,
      "grad_norm": 1.5526775121688843,
      "learning_rate": 2.3192423192423192e-06,
      "loss": 0.0264,
      "step": 3807
    },
    {
      "epoch": 3.9584199584199586,
      "grad_norm": 1.5784658193588257,
      "learning_rate": 2.316932316932317e-06,
      "loss": 0.0316,
      "step": 3808
    },
    {
      "epoch": 3.9594594594594597,
      "grad_norm": 2.6113502979278564,
      "learning_rate": 2.3146223146223148e-06,
      "loss": 0.1145,
      "step": 3809
    },
    {
      "epoch": 3.9604989604989607,
      "grad_norm": 3.468583822250366,
      "learning_rate": 2.3123123123123125e-06,
      "loss": 0.1322,
      "step": 3810
    },
    {
      "epoch": 3.9615384615384617,
      "grad_norm": 0.004322944208979607,
      "learning_rate": 2.3100023100023103e-06,
      "loss": 0.0001,
      "step": 3811
    },
    {
      "epoch": 3.9625779625779627,
      "grad_norm": 0.10281362384557724,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.0021,
      "step": 3812
    },
    {
      "epoch": 3.9636174636174637,
      "grad_norm": 0.28737759590148926,
      "learning_rate": 2.3053823053823054e-06,
      "loss": 0.0046,
      "step": 3813
    },
    {
      "epoch": 3.9646569646569647,
      "grad_norm": 14.735072135925293,
      "learning_rate": 2.303072303072303e-06,
      "loss": 1.09,
      "step": 3814
    },
    {
      "epoch": 3.9656964656964657,
      "grad_norm": 19.965375900268555,
      "learning_rate": 2.300762300762301e-06,
      "loss": 0.5098,
      "step": 3815
    },
    {
      "epoch": 3.9667359667359667,
      "grad_norm": 0.02133842557668686,
      "learning_rate": 2.2984522984522987e-06,
      "loss": 0.0006,
      "step": 3816
    },
    {
      "epoch": 3.9677754677754677,
      "grad_norm": 14.559213638305664,
      "learning_rate": 2.2961422961422965e-06,
      "loss": 1.8074,
      "step": 3817
    },
    {
      "epoch": 3.9688149688149688,
      "grad_norm": 0.0009462753660045564,
      "learning_rate": 2.2938322938322943e-06,
      "loss": 0.0,
      "step": 3818
    },
    {
      "epoch": 3.9698544698544698,
      "grad_norm": 1.4533599615097046,
      "learning_rate": 2.2915222915222916e-06,
      "loss": 0.0346,
      "step": 3819
    },
    {
      "epoch": 3.970893970893971,
      "grad_norm": 2.8870737552642822,
      "learning_rate": 2.2892122892122894e-06,
      "loss": 0.0598,
      "step": 3820
    },
    {
      "epoch": 3.971933471933472,
      "grad_norm": 4.132802963256836,
      "learning_rate": 2.286902286902287e-06,
      "loss": 0.3874,
      "step": 3821
    },
    {
      "epoch": 3.972972972972973,
      "grad_norm": 8.681702613830566,
      "learning_rate": 2.2845922845922845e-06,
      "loss": 0.3768,
      "step": 3822
    },
    {
      "epoch": 3.974012474012474,
      "grad_norm": 0.6327468156814575,
      "learning_rate": 2.2822822822822822e-06,
      "loss": 0.0134,
      "step": 3823
    },
    {
      "epoch": 3.975051975051975,
      "grad_norm": 2.971172332763672,
      "learning_rate": 2.27997227997228e-06,
      "loss": 0.0713,
      "step": 3824
    },
    {
      "epoch": 3.976091476091476,
      "grad_norm": 8.309932708740234,
      "learning_rate": 2.2776622776622778e-06,
      "loss": 0.5183,
      "step": 3825
    },
    {
      "epoch": 3.9771309771309773,
      "grad_norm": 13.4599027633667,
      "learning_rate": 2.2753522753522755e-06,
      "loss": 1.2856,
      "step": 3826
    },
    {
      "epoch": 3.9781704781704783,
      "grad_norm": 11.234210014343262,
      "learning_rate": 2.2730422730422733e-06,
      "loss": 0.8115,
      "step": 3827
    },
    {
      "epoch": 3.9792099792099793,
      "grad_norm": 0.08762773871421814,
      "learning_rate": 2.270732270732271e-06,
      "loss": 0.0017,
      "step": 3828
    },
    {
      "epoch": 3.9802494802494803,
      "grad_norm": 5.529712200164795,
      "learning_rate": 2.268422268422269e-06,
      "loss": 0.509,
      "step": 3829
    },
    {
      "epoch": 3.9812889812889813,
      "grad_norm": 2.618068218231201,
      "learning_rate": 2.2661122661122666e-06,
      "loss": 0.0958,
      "step": 3830
    },
    {
      "epoch": 3.9823284823284824,
      "grad_norm": 8.357830047607422,
      "learning_rate": 2.263802263802264e-06,
      "loss": 0.2174,
      "step": 3831
    },
    {
      "epoch": 3.9833679833679834,
      "grad_norm": 4.5352277755737305,
      "learning_rate": 2.2614922614922617e-06,
      "loss": 0.1564,
      "step": 3832
    },
    {
      "epoch": 3.9844074844074844,
      "grad_norm": 0.011746260337531567,
      "learning_rate": 2.2591822591822595e-06,
      "loss": 0.0002,
      "step": 3833
    },
    {
      "epoch": 3.9854469854469854,
      "grad_norm": 0.48385775089263916,
      "learning_rate": 2.256872256872257e-06,
      "loss": 0.0127,
      "step": 3834
    },
    {
      "epoch": 3.9864864864864864,
      "grad_norm": 5.331934452056885,
      "learning_rate": 2.2545622545622546e-06,
      "loss": 0.4049,
      "step": 3835
    },
    {
      "epoch": 3.9875259875259874,
      "grad_norm": 0.004386260639876127,
      "learning_rate": 2.2522522522522524e-06,
      "loss": 0.0001,
      "step": 3836
    },
    {
      "epoch": 3.9885654885654884,
      "grad_norm": 24.71803092956543,
      "learning_rate": 2.24994224994225e-06,
      "loss": 0.0714,
      "step": 3837
    },
    {
      "epoch": 3.98960498960499,
      "grad_norm": 8.399441719055176,
      "learning_rate": 2.247632247632248e-06,
      "loss": 0.4843,
      "step": 3838
    },
    {
      "epoch": 3.990644490644491,
      "grad_norm": 6.994706153869629,
      "learning_rate": 2.2453222453222457e-06,
      "loss": 0.591,
      "step": 3839
    },
    {
      "epoch": 3.991683991683992,
      "grad_norm": 0.43649402260780334,
      "learning_rate": 2.243012243012243e-06,
      "loss": 0.0052,
      "step": 3840
    },
    {
      "epoch": 3.992723492723493,
      "grad_norm": 1.914015769958496,
      "learning_rate": 2.2407022407022408e-06,
      "loss": 0.074,
      "step": 3841
    },
    {
      "epoch": 3.993762993762994,
      "grad_norm": 0.054117415100336075,
      "learning_rate": 2.2383922383922386e-06,
      "loss": 0.001,
      "step": 3842
    },
    {
      "epoch": 3.994802494802495,
      "grad_norm": 3.9509143829345703,
      "learning_rate": 2.2360822360822363e-06,
      "loss": 0.0723,
      "step": 3843
    },
    {
      "epoch": 3.995841995841996,
      "grad_norm": 12.29439640045166,
      "learning_rate": 2.233772233772234e-06,
      "loss": 0.9938,
      "step": 3844
    },
    {
      "epoch": 3.996881496881497,
      "grad_norm": 0.023738661780953407,
      "learning_rate": 2.231462231462232e-06,
      "loss": 0.0005,
      "step": 3845
    },
    {
      "epoch": 3.997920997920998,
      "grad_norm": 8.44709587097168,
      "learning_rate": 2.229152229152229e-06,
      "loss": 0.3122,
      "step": 3846
    },
    {
      "epoch": 3.998960498960499,
      "grad_norm": 11.878024101257324,
      "learning_rate": 2.226842226842227e-06,
      "loss": 0.8965,
      "step": 3847
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.6288987398147583,
      "learning_rate": 2.2245322245322247e-06,
      "loss": 0.0147,
      "step": 3848
    },
    {
      "epoch": 4.001039501039501,
      "grad_norm": 4.34573221206665,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.299,
      "step": 3849
    },
    {
      "epoch": 4.002079002079002,
      "grad_norm": 0.3988482654094696,
      "learning_rate": 2.21991221991222e-06,
      "loss": 0.0088,
      "step": 3850
    },
    {
      "epoch": 4.003118503118503,
      "grad_norm": 5.311618328094482,
      "learning_rate": 2.2176022176022176e-06,
      "loss": 0.1288,
      "step": 3851
    },
    {
      "epoch": 4.004158004158004,
      "grad_norm": 0.03252868354320526,
      "learning_rate": 2.2152922152922154e-06,
      "loss": 0.0008,
      "step": 3852
    },
    {
      "epoch": 4.005197505197505,
      "grad_norm": 1.65116286277771,
      "learning_rate": 2.212982212982213e-06,
      "loss": 0.0598,
      "step": 3853
    },
    {
      "epoch": 4.006237006237006,
      "grad_norm": 14.985928535461426,
      "learning_rate": 2.210672210672211e-06,
      "loss": 0.6218,
      "step": 3854
    },
    {
      "epoch": 4.007276507276507,
      "grad_norm": 9.496588706970215,
      "learning_rate": 2.2083622083622087e-06,
      "loss": 0.7732,
      "step": 3855
    },
    {
      "epoch": 4.008316008316008,
      "grad_norm": 7.977664947509766,
      "learning_rate": 2.2060522060522064e-06,
      "loss": 0.7576,
      "step": 3856
    },
    {
      "epoch": 4.009355509355509,
      "grad_norm": 6.168428421020508,
      "learning_rate": 2.2037422037422042e-06,
      "loss": 0.3512,
      "step": 3857
    },
    {
      "epoch": 4.01039501039501,
      "grad_norm": 7.742276668548584,
      "learning_rate": 2.2014322014322016e-06,
      "loss": 0.1884,
      "step": 3858
    },
    {
      "epoch": 4.011434511434511,
      "grad_norm": 0.12430943548679352,
      "learning_rate": 2.1991221991221993e-06,
      "loss": 0.0033,
      "step": 3859
    },
    {
      "epoch": 4.012474012474012,
      "grad_norm": 3.1924452781677246,
      "learning_rate": 2.196812196812197e-06,
      "loss": 0.0411,
      "step": 3860
    },
    {
      "epoch": 4.013513513513513,
      "grad_norm": 2.041194200515747,
      "learning_rate": 2.194502194502195e-06,
      "loss": 0.0927,
      "step": 3861
    },
    {
      "epoch": 4.014553014553014,
      "grad_norm": 0.478383332490921,
      "learning_rate": 2.192192192192192e-06,
      "loss": 0.0099,
      "step": 3862
    },
    {
      "epoch": 4.015592515592515,
      "grad_norm": 8.808903694152832,
      "learning_rate": 2.18988218988219e-06,
      "loss": 0.411,
      "step": 3863
    },
    {
      "epoch": 4.016632016632016,
      "grad_norm": 4.590777397155762,
      "learning_rate": 2.1875721875721877e-06,
      "loss": 0.0937,
      "step": 3864
    },
    {
      "epoch": 4.017671517671518,
      "grad_norm": 0.27158844470977783,
      "learning_rate": 2.1852621852621855e-06,
      "loss": 0.0058,
      "step": 3865
    },
    {
      "epoch": 4.018711018711019,
      "grad_norm": 6.849306106567383,
      "learning_rate": 2.1829521829521833e-06,
      "loss": 0.4817,
      "step": 3866
    },
    {
      "epoch": 4.01975051975052,
      "grad_norm": 2.9405770301818848,
      "learning_rate": 2.1806421806421806e-06,
      "loss": 0.0536,
      "step": 3867
    },
    {
      "epoch": 4.020790020790021,
      "grad_norm": 3.641545057296753,
      "learning_rate": 2.1783321783321784e-06,
      "loss": 0.1562,
      "step": 3868
    },
    {
      "epoch": 4.021829521829522,
      "grad_norm": 1.2189357280731201,
      "learning_rate": 2.176022176022176e-06,
      "loss": 0.029,
      "step": 3869
    },
    {
      "epoch": 4.022869022869023,
      "grad_norm": 0.2723861038684845,
      "learning_rate": 2.173712173712174e-06,
      "loss": 0.0076,
      "step": 3870
    },
    {
      "epoch": 4.023908523908524,
      "grad_norm": 3.9028191566467285,
      "learning_rate": 2.1714021714021717e-06,
      "loss": 0.1592,
      "step": 3871
    },
    {
      "epoch": 4.024948024948025,
      "grad_norm": 5.267368316650391,
      "learning_rate": 2.1690921690921694e-06,
      "loss": 0.1642,
      "step": 3872
    },
    {
      "epoch": 4.025987525987526,
      "grad_norm": 2.9735355377197266,
      "learning_rate": 2.1667821667821672e-06,
      "loss": 0.061,
      "step": 3873
    },
    {
      "epoch": 4.027027027027027,
      "grad_norm": 1.855345368385315,
      "learning_rate": 2.1644721644721646e-06,
      "loss": 0.0525,
      "step": 3874
    },
    {
      "epoch": 4.028066528066528,
      "grad_norm": 8.544869422912598,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 0.6427,
      "step": 3875
    },
    {
      "epoch": 4.029106029106029,
      "grad_norm": 9.181122779846191,
      "learning_rate": 2.15985215985216e-06,
      "loss": 0.5065,
      "step": 3876
    },
    {
      "epoch": 4.03014553014553,
      "grad_norm": 11.528375625610352,
      "learning_rate": 2.1575421575421574e-06,
      "loss": 0.3951,
      "step": 3877
    },
    {
      "epoch": 4.031185031185031,
      "grad_norm": 0.14364181458950043,
      "learning_rate": 2.155232155232155e-06,
      "loss": 0.0039,
      "step": 3878
    },
    {
      "epoch": 4.032224532224532,
      "grad_norm": 10.91196060180664,
      "learning_rate": 2.152922152922153e-06,
      "loss": 0.5188,
      "step": 3879
    },
    {
      "epoch": 4.033264033264033,
      "grad_norm": 3.6201767921447754,
      "learning_rate": 2.1506121506121507e-06,
      "loss": 0.3075,
      "step": 3880
    },
    {
      "epoch": 4.034303534303534,
      "grad_norm": 1.5775871276855469,
      "learning_rate": 2.1483021483021485e-06,
      "loss": 0.0861,
      "step": 3881
    },
    {
      "epoch": 4.035343035343035,
      "grad_norm": 0.17233553528785706,
      "learning_rate": 2.1459921459921463e-06,
      "loss": 0.0046,
      "step": 3882
    },
    {
      "epoch": 4.036382536382536,
      "grad_norm": 1.6797047853469849,
      "learning_rate": 2.143682143682144e-06,
      "loss": 0.0766,
      "step": 3883
    },
    {
      "epoch": 4.037422037422037,
      "grad_norm": 0.03757861256599426,
      "learning_rate": 2.141372141372142e-06,
      "loss": 0.0012,
      "step": 3884
    },
    {
      "epoch": 4.038461538461538,
      "grad_norm": 6.65224552154541,
      "learning_rate": 2.139062139062139e-06,
      "loss": 0.3537,
      "step": 3885
    },
    {
      "epoch": 4.039501039501039,
      "grad_norm": 2.8057408332824707,
      "learning_rate": 2.136752136752137e-06,
      "loss": 0.1392,
      "step": 3886
    },
    {
      "epoch": 4.04054054054054,
      "grad_norm": 13.577767372131348,
      "learning_rate": 2.1344421344421347e-06,
      "loss": 0.825,
      "step": 3887
    },
    {
      "epoch": 4.041580041580041,
      "grad_norm": 21.758955001831055,
      "learning_rate": 2.1321321321321325e-06,
      "loss": 1.6505,
      "step": 3888
    },
    {
      "epoch": 4.042619542619542,
      "grad_norm": 0.7290350198745728,
      "learning_rate": 2.12982212982213e-06,
      "loss": 0.0148,
      "step": 3889
    },
    {
      "epoch": 4.043659043659043,
      "grad_norm": 4.572800159454346,
      "learning_rate": 2.1275121275121276e-06,
      "loss": 0.3039,
      "step": 3890
    },
    {
      "epoch": 4.044698544698544,
      "grad_norm": 1.6838501691818237,
      "learning_rate": 2.1252021252021253e-06,
      "loss": 0.2458,
      "step": 3891
    },
    {
      "epoch": 4.045738045738045,
      "grad_norm": 0.09158274531364441,
      "learning_rate": 2.122892122892123e-06,
      "loss": 0.0022,
      "step": 3892
    },
    {
      "epoch": 4.046777546777546,
      "grad_norm": 1.051876425743103,
      "learning_rate": 2.120582120582121e-06,
      "loss": 0.0301,
      "step": 3893
    },
    {
      "epoch": 4.047817047817047,
      "grad_norm": 7.565310955047607,
      "learning_rate": 2.118272118272118e-06,
      "loss": 0.5889,
      "step": 3894
    },
    {
      "epoch": 4.048856548856548,
      "grad_norm": 0.4519527852535248,
      "learning_rate": 2.115962115962116e-06,
      "loss": 0.0051,
      "step": 3895
    },
    {
      "epoch": 4.04989604989605,
      "grad_norm": 11.864056587219238,
      "learning_rate": 2.1136521136521137e-06,
      "loss": 0.6653,
      "step": 3896
    },
    {
      "epoch": 4.050935550935551,
      "grad_norm": 3.6035194396972656,
      "learning_rate": 2.1113421113421115e-06,
      "loss": 0.0504,
      "step": 3897
    },
    {
      "epoch": 4.051975051975052,
      "grad_norm": 2.7916107177734375,
      "learning_rate": 2.1090321090321093e-06,
      "loss": 0.0709,
      "step": 3898
    },
    {
      "epoch": 4.053014553014553,
      "grad_norm": 0.2317170947790146,
      "learning_rate": 2.106722106722107e-06,
      "loss": 0.0072,
      "step": 3899
    },
    {
      "epoch": 4.054054054054054,
      "grad_norm": 3.6809520721435547,
      "learning_rate": 2.104412104412105e-06,
      "loss": 0.0592,
      "step": 3900
    },
    {
      "epoch": 4.055093555093555,
      "grad_norm": 12.056526184082031,
      "learning_rate": 2.102102102102102e-06,
      "loss": 1.1791,
      "step": 3901
    },
    {
      "epoch": 4.056133056133056,
      "grad_norm": 0.6267997622489929,
      "learning_rate": 2.0997920997921e-06,
      "loss": 0.026,
      "step": 3902
    },
    {
      "epoch": 4.057172557172557,
      "grad_norm": 7.745768070220947,
      "learning_rate": 2.0974820974820977e-06,
      "loss": 0.1447,
      "step": 3903
    },
    {
      "epoch": 4.058212058212058,
      "grad_norm": 0.046626631170511246,
      "learning_rate": 2.095172095172095e-06,
      "loss": 0.0009,
      "step": 3904
    },
    {
      "epoch": 4.0592515592515594,
      "grad_norm": 0.03431563451886177,
      "learning_rate": 2.092862092862093e-06,
      "loss": 0.001,
      "step": 3905
    },
    {
      "epoch": 4.0602910602910605,
      "grad_norm": 0.21163296699523926,
      "learning_rate": 2.0905520905520906e-06,
      "loss": 0.0064,
      "step": 3906
    },
    {
      "epoch": 4.0613305613305615,
      "grad_norm": 2.4047062397003174,
      "learning_rate": 2.0882420882420883e-06,
      "loss": 0.0353,
      "step": 3907
    },
    {
      "epoch": 4.0623700623700625,
      "grad_norm": 7.710663318634033,
      "learning_rate": 2.085932085932086e-06,
      "loss": 0.5675,
      "step": 3908
    },
    {
      "epoch": 4.0634095634095635,
      "grad_norm": 3.3957293033599854,
      "learning_rate": 2.083622083622084e-06,
      "loss": 0.1848,
      "step": 3909
    },
    {
      "epoch": 4.0644490644490645,
      "grad_norm": 8.47187328338623,
      "learning_rate": 2.0813120813120816e-06,
      "loss": 0.1675,
      "step": 3910
    },
    {
      "epoch": 4.0654885654885655,
      "grad_norm": 0.38779881596565247,
      "learning_rate": 2.0790020790020794e-06,
      "loss": 0.0084,
      "step": 3911
    },
    {
      "epoch": 4.0665280665280665,
      "grad_norm": 11.849868774414062,
      "learning_rate": 2.0766920766920767e-06,
      "loss": 1.1938,
      "step": 3912
    },
    {
      "epoch": 4.0675675675675675,
      "grad_norm": 7.814253330230713,
      "learning_rate": 2.0743820743820745e-06,
      "loss": 0.3056,
      "step": 3913
    },
    {
      "epoch": 4.0686070686070686,
      "grad_norm": 9.556112289428711,
      "learning_rate": 2.0720720720720723e-06,
      "loss": 0.4659,
      "step": 3914
    },
    {
      "epoch": 4.06964656964657,
      "grad_norm": 0.10459741204977036,
      "learning_rate": 2.06976206976207e-06,
      "loss": 0.0019,
      "step": 3915
    },
    {
      "epoch": 4.070686070686071,
      "grad_norm": 2.1530704498291016,
      "learning_rate": 2.067452067452068e-06,
      "loss": 0.0306,
      "step": 3916
    },
    {
      "epoch": 4.071725571725572,
      "grad_norm": 9.012895584106445,
      "learning_rate": 2.065142065142065e-06,
      "loss": 0.4619,
      "step": 3917
    },
    {
      "epoch": 4.072765072765073,
      "grad_norm": 10.304547309875488,
      "learning_rate": 2.062832062832063e-06,
      "loss": 0.4175,
      "step": 3918
    },
    {
      "epoch": 4.073804573804574,
      "grad_norm": 8.123601913452148,
      "learning_rate": 2.0605220605220607e-06,
      "loss": 0.4888,
      "step": 3919
    },
    {
      "epoch": 4.074844074844075,
      "grad_norm": 0.0011895884526893497,
      "learning_rate": 2.0582120582120585e-06,
      "loss": 0.0,
      "step": 3920
    },
    {
      "epoch": 4.075883575883576,
      "grad_norm": 1.604593276977539,
      "learning_rate": 2.055902055902056e-06,
      "loss": 0.0495,
      "step": 3921
    },
    {
      "epoch": 4.076923076923077,
      "grad_norm": 3.2594895362854004,
      "learning_rate": 2.0535920535920536e-06,
      "loss": 0.0281,
      "step": 3922
    },
    {
      "epoch": 4.077962577962578,
      "grad_norm": 0.15859843790531158,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 0.004,
      "step": 3923
    },
    {
      "epoch": 4.079002079002079,
      "grad_norm": 0.02080628275871277,
      "learning_rate": 2.048972048972049e-06,
      "loss": 0.0004,
      "step": 3924
    },
    {
      "epoch": 4.08004158004158,
      "grad_norm": 0.42855557799339294,
      "learning_rate": 2.046662046662047e-06,
      "loss": 0.0045,
      "step": 3925
    },
    {
      "epoch": 4.081081081081081,
      "grad_norm": 4.763267517089844,
      "learning_rate": 2.0443520443520446e-06,
      "loss": 0.252,
      "step": 3926
    },
    {
      "epoch": 4.082120582120582,
      "grad_norm": 7.961658477783203,
      "learning_rate": 2.0420420420420424e-06,
      "loss": 0.4169,
      "step": 3927
    },
    {
      "epoch": 4.083160083160083,
      "grad_norm": 12.200848579406738,
      "learning_rate": 2.03973203973204e-06,
      "loss": 0.5576,
      "step": 3928
    },
    {
      "epoch": 4.084199584199585,
      "grad_norm": 0.2671249210834503,
      "learning_rate": 2.0374220374220375e-06,
      "loss": 0.0063,
      "step": 3929
    },
    {
      "epoch": 4.085239085239086,
      "grad_norm": 2.368168592453003,
      "learning_rate": 2.0351120351120353e-06,
      "loss": 0.0575,
      "step": 3930
    },
    {
      "epoch": 4.086278586278587,
      "grad_norm": 10.609644889831543,
      "learning_rate": 2.032802032802033e-06,
      "loss": 0.4751,
      "step": 3931
    },
    {
      "epoch": 4.087318087318088,
      "grad_norm": 7.265843391418457,
      "learning_rate": 2.0304920304920304e-06,
      "loss": 0.4997,
      "step": 3932
    },
    {
      "epoch": 4.088357588357589,
      "grad_norm": 10.27369213104248,
      "learning_rate": 2.028182028182028e-06,
      "loss": 0.3908,
      "step": 3933
    },
    {
      "epoch": 4.08939708939709,
      "grad_norm": 2.5010275840759277,
      "learning_rate": 2.025872025872026e-06,
      "loss": 0.2932,
      "step": 3934
    },
    {
      "epoch": 4.090436590436591,
      "grad_norm": 1.6605794429779053,
      "learning_rate": 2.0235620235620237e-06,
      "loss": 0.0376,
      "step": 3935
    },
    {
      "epoch": 4.091476091476092,
      "grad_norm": 4.947322845458984,
      "learning_rate": 2.0212520212520215e-06,
      "loss": 0.1477,
      "step": 3936
    },
    {
      "epoch": 4.092515592515593,
      "grad_norm": 0.015340354293584824,
      "learning_rate": 2.0189420189420192e-06,
      "loss": 0.0003,
      "step": 3937
    },
    {
      "epoch": 4.093555093555094,
      "grad_norm": 12.941651344299316,
      "learning_rate": 2.016632016632017e-06,
      "loss": 0.6778,
      "step": 3938
    },
    {
      "epoch": 4.094594594594595,
      "grad_norm": 9.275856971740723,
      "learning_rate": 2.0143220143220143e-06,
      "loss": 0.741,
      "step": 3939
    },
    {
      "epoch": 4.095634095634096,
      "grad_norm": 2.234531879425049,
      "learning_rate": 2.012012012012012e-06,
      "loss": 0.3126,
      "step": 3940
    },
    {
      "epoch": 4.096673596673597,
      "grad_norm": 4.1659369468688965,
      "learning_rate": 2.00970200970201e-06,
      "loss": 0.1181,
      "step": 3941
    },
    {
      "epoch": 4.097713097713098,
      "grad_norm": 10.8466215133667,
      "learning_rate": 2.0073920073920076e-06,
      "loss": 0.5613,
      "step": 3942
    },
    {
      "epoch": 4.098752598752599,
      "grad_norm": 1.559237003326416,
      "learning_rate": 2.0050820050820054e-06,
      "loss": 0.0277,
      "step": 3943
    },
    {
      "epoch": 4.0997920997921,
      "grad_norm": 7.026435375213623,
      "learning_rate": 2.0027720027720028e-06,
      "loss": 0.2141,
      "step": 3944
    },
    {
      "epoch": 4.100831600831601,
      "grad_norm": 6.664749622344971,
      "learning_rate": 2.0004620004620005e-06,
      "loss": 0.2036,
      "step": 3945
    },
    {
      "epoch": 4.101871101871102,
      "grad_norm": 2.792083501815796,
      "learning_rate": 1.9981519981519983e-06,
      "loss": 0.0606,
      "step": 3946
    },
    {
      "epoch": 4.102910602910603,
      "grad_norm": 6.02000093460083,
      "learning_rate": 1.995841995841996e-06,
      "loss": 0.1256,
      "step": 3947
    },
    {
      "epoch": 4.103950103950104,
      "grad_norm": 4.715527057647705,
      "learning_rate": 1.9935319935319934e-06,
      "loss": 0.1299,
      "step": 3948
    },
    {
      "epoch": 4.104989604989605,
      "grad_norm": 0.0066122873686254025,
      "learning_rate": 1.991221991221991e-06,
      "loss": 0.0001,
      "step": 3949
    },
    {
      "epoch": 4.106029106029106,
      "grad_norm": 3.474184274673462,
      "learning_rate": 1.988911988911989e-06,
      "loss": 0.2328,
      "step": 3950
    },
    {
      "epoch": 4.107068607068607,
      "grad_norm": 4.632596015930176,
      "learning_rate": 1.9866019866019867e-06,
      "loss": 0.1278,
      "step": 3951
    },
    {
      "epoch": 4.108108108108108,
      "grad_norm": 3.1452908515930176,
      "learning_rate": 1.9842919842919845e-06,
      "loss": 0.061,
      "step": 3952
    },
    {
      "epoch": 4.109147609147609,
      "grad_norm": 0.2267417013645172,
      "learning_rate": 1.9819819819819822e-06,
      "loss": 0.0059,
      "step": 3953
    },
    {
      "epoch": 4.11018711018711,
      "grad_norm": 1.7719359397888184,
      "learning_rate": 1.97967197967198e-06,
      "loss": 0.07,
      "step": 3954
    },
    {
      "epoch": 4.111226611226611,
      "grad_norm": 1.1216524839401245,
      "learning_rate": 1.9773619773619778e-06,
      "loss": 0.0213,
      "step": 3955
    },
    {
      "epoch": 4.112266112266112,
      "grad_norm": 5.114573001861572,
      "learning_rate": 1.975051975051975e-06,
      "loss": 0.1357,
      "step": 3956
    },
    {
      "epoch": 4.113305613305613,
      "grad_norm": 9.502123832702637,
      "learning_rate": 1.972741972741973e-06,
      "loss": 0.7736,
      "step": 3957
    },
    {
      "epoch": 4.114345114345114,
      "grad_norm": 6.611077308654785,
      "learning_rate": 1.9704319704319707e-06,
      "loss": 0.1888,
      "step": 3958
    },
    {
      "epoch": 4.115384615384615,
      "grad_norm": 3.405991792678833,
      "learning_rate": 1.968121968121968e-06,
      "loss": 0.0855,
      "step": 3959
    },
    {
      "epoch": 4.116424116424117,
      "grad_norm": 0.3084595501422882,
      "learning_rate": 1.9658119658119658e-06,
      "loss": 0.0088,
      "step": 3960
    },
    {
      "epoch": 4.117463617463618,
      "grad_norm": 0.11557632684707642,
      "learning_rate": 1.9635019635019635e-06,
      "loss": 0.0023,
      "step": 3961
    },
    {
      "epoch": 4.118503118503119,
      "grad_norm": 0.4615918695926666,
      "learning_rate": 1.9611919611919613e-06,
      "loss": 0.0136,
      "step": 3962
    },
    {
      "epoch": 4.11954261954262,
      "grad_norm": 8.216227531433105,
      "learning_rate": 1.958881958881959e-06,
      "loss": 0.5017,
      "step": 3963
    },
    {
      "epoch": 4.120582120582121,
      "grad_norm": 1.2277436256408691,
      "learning_rate": 1.956571956571957e-06,
      "loss": 0.0259,
      "step": 3964
    },
    {
      "epoch": 4.121621621621622,
      "grad_norm": 11.282058715820312,
      "learning_rate": 1.9542619542619546e-06,
      "loss": 0.7121,
      "step": 3965
    },
    {
      "epoch": 4.122661122661123,
      "grad_norm": 0.6728746294975281,
      "learning_rate": 1.951951951951952e-06,
      "loss": 0.02,
      "step": 3966
    },
    {
      "epoch": 4.123700623700624,
      "grad_norm": 0.007000439800322056,
      "learning_rate": 1.9496419496419497e-06,
      "loss": 0.0002,
      "step": 3967
    },
    {
      "epoch": 4.124740124740125,
      "grad_norm": 8.943163871765137,
      "learning_rate": 1.9473319473319475e-06,
      "loss": 0.5562,
      "step": 3968
    },
    {
      "epoch": 4.125779625779626,
      "grad_norm": 11.286050796508789,
      "learning_rate": 1.9450219450219452e-06,
      "loss": 0.7542,
      "step": 3969
    },
    {
      "epoch": 4.126819126819127,
      "grad_norm": 0.1594606339931488,
      "learning_rate": 1.942711942711943e-06,
      "loss": 0.0057,
      "step": 3970
    },
    {
      "epoch": 4.127858627858628,
      "grad_norm": 3.0805113315582275,
      "learning_rate": 1.9404019404019408e-06,
      "loss": 0.1091,
      "step": 3971
    },
    {
      "epoch": 4.128898128898129,
      "grad_norm": 0.0031344450544565916,
      "learning_rate": 1.938091938091938e-06,
      "loss": 0.0001,
      "step": 3972
    },
    {
      "epoch": 4.12993762993763,
      "grad_norm": 5.24778938293457,
      "learning_rate": 1.935781935781936e-06,
      "loss": 0.1809,
      "step": 3973
    },
    {
      "epoch": 4.130977130977131,
      "grad_norm": 0.20770363509655,
      "learning_rate": 1.9334719334719337e-06,
      "loss": 0.0038,
      "step": 3974
    },
    {
      "epoch": 4.132016632016632,
      "grad_norm": 0.0760016217827797,
      "learning_rate": 1.9311619311619314e-06,
      "loss": 0.0012,
      "step": 3975
    },
    {
      "epoch": 4.133056133056133,
      "grad_norm": 1.8212653398513794,
      "learning_rate": 1.9288519288519288e-06,
      "loss": 0.0574,
      "step": 3976
    },
    {
      "epoch": 4.134095634095634,
      "grad_norm": 3.7411952018737793,
      "learning_rate": 1.9265419265419265e-06,
      "loss": 0.3133,
      "step": 3977
    },
    {
      "epoch": 4.135135135135135,
      "grad_norm": 8.470130920410156,
      "learning_rate": 1.9242319242319243e-06,
      "loss": 0.1549,
      "step": 3978
    },
    {
      "epoch": 4.136174636174636,
      "grad_norm": 13.471936225891113,
      "learning_rate": 1.921921921921922e-06,
      "loss": 1.9759,
      "step": 3979
    },
    {
      "epoch": 4.137214137214137,
      "grad_norm": 10.0439453125,
      "learning_rate": 1.91961191961192e-06,
      "loss": 0.5779,
      "step": 3980
    },
    {
      "epoch": 4.138253638253638,
      "grad_norm": 0.3061752915382385,
      "learning_rate": 1.9173019173019176e-06,
      "loss": 0.008,
      "step": 3981
    },
    {
      "epoch": 4.139293139293139,
      "grad_norm": 0.16787175834178925,
      "learning_rate": 1.9149919149919154e-06,
      "loss": 0.0025,
      "step": 3982
    },
    {
      "epoch": 4.14033264033264,
      "grad_norm": 0.1021217480301857,
      "learning_rate": 1.912681912681913e-06,
      "loss": 0.0022,
      "step": 3983
    },
    {
      "epoch": 4.141372141372141,
      "grad_norm": 1.6589683294296265,
      "learning_rate": 1.9103719103719105e-06,
      "loss": 0.046,
      "step": 3984
    },
    {
      "epoch": 4.142411642411642,
      "grad_norm": 4.064008712768555,
      "learning_rate": 1.9080619080619082e-06,
      "loss": 0.0336,
      "step": 3985
    },
    {
      "epoch": 4.143451143451143,
      "grad_norm": 0.7545316815376282,
      "learning_rate": 1.9057519057519058e-06,
      "loss": 0.0158,
      "step": 3986
    },
    {
      "epoch": 4.144490644490644,
      "grad_norm": 1.8117564916610718,
      "learning_rate": 1.9034419034419036e-06,
      "loss": 0.4961,
      "step": 3987
    },
    {
      "epoch": 4.145530145530145,
      "grad_norm": 0.3270220160484314,
      "learning_rate": 1.9011319011319013e-06,
      "loss": 0.0122,
      "step": 3988
    },
    {
      "epoch": 4.146569646569646,
      "grad_norm": 5.634047031402588,
      "learning_rate": 1.8988218988218989e-06,
      "loss": 0.3619,
      "step": 3989
    },
    {
      "epoch": 4.147609147609147,
      "grad_norm": 0.004145979881286621,
      "learning_rate": 1.8965118965118967e-06,
      "loss": 0.0001,
      "step": 3990
    },
    {
      "epoch": 4.148648648648648,
      "grad_norm": 2.454077959060669,
      "learning_rate": 1.8942018942018944e-06,
      "loss": 0.0743,
      "step": 3991
    },
    {
      "epoch": 4.149688149688149,
      "grad_norm": 2.424099922180176,
      "learning_rate": 1.8918918918918922e-06,
      "loss": 0.0716,
      "step": 3992
    },
    {
      "epoch": 4.150727650727651,
      "grad_norm": 0.14816658198833466,
      "learning_rate": 1.88958188958189e-06,
      "loss": 0.0032,
      "step": 3993
    },
    {
      "epoch": 4.151767151767152,
      "grad_norm": 9.936027526855469,
      "learning_rate": 1.8872718872718873e-06,
      "loss": 0.9768,
      "step": 3994
    },
    {
      "epoch": 4.152806652806653,
      "grad_norm": 7.472232818603516,
      "learning_rate": 1.884961884961885e-06,
      "loss": 0.2945,
      "step": 3995
    },
    {
      "epoch": 4.153846153846154,
      "grad_norm": 5.426685810089111,
      "learning_rate": 1.8826518826518828e-06,
      "loss": 0.2832,
      "step": 3996
    },
    {
      "epoch": 4.154885654885655,
      "grad_norm": 2.724226951599121,
      "learning_rate": 1.8803418803418804e-06,
      "loss": 0.0439,
      "step": 3997
    },
    {
      "epoch": 4.155925155925156,
      "grad_norm": 1.9673157930374146,
      "learning_rate": 1.8780318780318782e-06,
      "loss": 0.0547,
      "step": 3998
    },
    {
      "epoch": 4.156964656964657,
      "grad_norm": 7.403589725494385,
      "learning_rate": 1.875721875721876e-06,
      "loss": 0.4418,
      "step": 3999
    },
    {
      "epoch": 4.158004158004158,
      "grad_norm": 7.950428009033203,
      "learning_rate": 1.8734118734118737e-06,
      "loss": 0.4021,
      "step": 4000
    },
    {
      "epoch": 4.159043659043659,
      "grad_norm": 9.602352142333984,
      "learning_rate": 1.8711018711018713e-06,
      "loss": 0.8566,
      "step": 4001
    },
    {
      "epoch": 4.16008316008316,
      "grad_norm": 1.825050711631775,
      "learning_rate": 1.868791868791869e-06,
      "loss": 0.026,
      "step": 4002
    },
    {
      "epoch": 4.161122661122661,
      "grad_norm": 7.144789218902588,
      "learning_rate": 1.8664818664818666e-06,
      "loss": 0.1889,
      "step": 4003
    },
    {
      "epoch": 4.162162162162162,
      "grad_norm": 2.014277458190918,
      "learning_rate": 1.8641718641718643e-06,
      "loss": 0.041,
      "step": 4004
    },
    {
      "epoch": 4.163201663201663,
      "grad_norm": 0.017938906326889992,
      "learning_rate": 1.861861861861862e-06,
      "loss": 0.0004,
      "step": 4005
    },
    {
      "epoch": 4.164241164241164,
      "grad_norm": 5.936906814575195,
      "learning_rate": 1.8595518595518597e-06,
      "loss": 0.1381,
      "step": 4006
    },
    {
      "epoch": 4.165280665280665,
      "grad_norm": 5.884034633636475,
      "learning_rate": 1.8572418572418574e-06,
      "loss": 0.4302,
      "step": 4007
    },
    {
      "epoch": 4.166320166320166,
      "grad_norm": 0.4456760883331299,
      "learning_rate": 1.8549318549318552e-06,
      "loss": 0.0147,
      "step": 4008
    },
    {
      "epoch": 4.167359667359667,
      "grad_norm": 4.608760356903076,
      "learning_rate": 1.8526218526218528e-06,
      "loss": 0.1292,
      "step": 4009
    },
    {
      "epoch": 4.168399168399168,
      "grad_norm": 8.948893547058105,
      "learning_rate": 1.8503118503118505e-06,
      "loss": 0.9863,
      "step": 4010
    },
    {
      "epoch": 4.169438669438669,
      "grad_norm": 0.216671884059906,
      "learning_rate": 1.8480018480018483e-06,
      "loss": 0.0044,
      "step": 4011
    },
    {
      "epoch": 4.17047817047817,
      "grad_norm": 4.625119209289551,
      "learning_rate": 1.8456918456918456e-06,
      "loss": 0.2856,
      "step": 4012
    },
    {
      "epoch": 4.171517671517671,
      "grad_norm": 0.2241864949464798,
      "learning_rate": 1.8433818433818434e-06,
      "loss": 0.0061,
      "step": 4013
    },
    {
      "epoch": 4.172557172557172,
      "grad_norm": 4.816661357879639,
      "learning_rate": 1.8410718410718412e-06,
      "loss": 0.5286,
      "step": 4014
    },
    {
      "epoch": 4.173596673596673,
      "grad_norm": 6.056797027587891,
      "learning_rate": 1.838761838761839e-06,
      "loss": 0.2242,
      "step": 4015
    },
    {
      "epoch": 4.174636174636174,
      "grad_norm": 1.5220786333084106,
      "learning_rate": 1.8364518364518367e-06,
      "loss": 0.0316,
      "step": 4016
    },
    {
      "epoch": 4.175675675675675,
      "grad_norm": 9.600642204284668,
      "learning_rate": 1.8341418341418343e-06,
      "loss": 0.6818,
      "step": 4017
    },
    {
      "epoch": 4.1767151767151764,
      "grad_norm": 0.1177779883146286,
      "learning_rate": 1.831831831831832e-06,
      "loss": 0.0037,
      "step": 4018
    },
    {
      "epoch": 4.1777546777546775,
      "grad_norm": 4.937497615814209,
      "learning_rate": 1.8295218295218298e-06,
      "loss": 0.3238,
      "step": 4019
    },
    {
      "epoch": 4.1787941787941785,
      "grad_norm": 5.371255874633789,
      "learning_rate": 1.8272118272118276e-06,
      "loss": 0.23,
      "step": 4020
    },
    {
      "epoch": 4.1798336798336795,
      "grad_norm": 1.4247781038284302,
      "learning_rate": 1.824901824901825e-06,
      "loss": 0.0328,
      "step": 4021
    },
    {
      "epoch": 4.1808731808731805,
      "grad_norm": 0.15081067383289337,
      "learning_rate": 1.8225918225918227e-06,
      "loss": 0.0032,
      "step": 4022
    },
    {
      "epoch": 4.1819126819126815,
      "grad_norm": 2.7095346450805664,
      "learning_rate": 1.8202818202818204e-06,
      "loss": 0.1122,
      "step": 4023
    },
    {
      "epoch": 4.182952182952183,
      "grad_norm": 1.09393310546875,
      "learning_rate": 1.817971817971818e-06,
      "loss": 0.0206,
      "step": 4024
    },
    {
      "epoch": 4.183991683991684,
      "grad_norm": 10.313549995422363,
      "learning_rate": 1.8156618156618158e-06,
      "loss": 0.8821,
      "step": 4025
    },
    {
      "epoch": 4.185031185031185,
      "grad_norm": 6.645352840423584,
      "learning_rate": 1.8133518133518135e-06,
      "loss": 0.4741,
      "step": 4026
    },
    {
      "epoch": 4.186070686070686,
      "grad_norm": 1.595228910446167,
      "learning_rate": 1.8110418110418113e-06,
      "loss": 0.0237,
      "step": 4027
    },
    {
      "epoch": 4.1871101871101875,
      "grad_norm": 4.622249126434326,
      "learning_rate": 1.808731808731809e-06,
      "loss": 0.1143,
      "step": 4028
    },
    {
      "epoch": 4.1881496881496885,
      "grad_norm": 18.4790096282959,
      "learning_rate": 1.8064218064218066e-06,
      "loss": 0.677,
      "step": 4029
    },
    {
      "epoch": 4.1891891891891895,
      "grad_norm": 7.102630615234375,
      "learning_rate": 1.8041118041118042e-06,
      "loss": 0.2497,
      "step": 4030
    },
    {
      "epoch": 4.1902286902286905,
      "grad_norm": 0.9145203828811646,
      "learning_rate": 1.801801801801802e-06,
      "loss": 0.0262,
      "step": 4031
    },
    {
      "epoch": 4.1912681912681915,
      "grad_norm": 0.6228851079940796,
      "learning_rate": 1.7994917994917995e-06,
      "loss": 0.0118,
      "step": 4032
    },
    {
      "epoch": 4.1923076923076925,
      "grad_norm": 0.08999498933553696,
      "learning_rate": 1.7971817971817973e-06,
      "loss": 0.0021,
      "step": 4033
    },
    {
      "epoch": 4.1933471933471935,
      "grad_norm": 1.8668020963668823,
      "learning_rate": 1.794871794871795e-06,
      "loss": 0.056,
      "step": 4034
    },
    {
      "epoch": 4.1943866943866945,
      "grad_norm": 0.015170056372880936,
      "learning_rate": 1.7925617925617928e-06,
      "loss": 0.0003,
      "step": 4035
    },
    {
      "epoch": 4.1954261954261955,
      "grad_norm": 3.824902057647705,
      "learning_rate": 1.7902517902517906e-06,
      "loss": 0.1182,
      "step": 4036
    },
    {
      "epoch": 4.196465696465697,
      "grad_norm": 4.108384609222412,
      "learning_rate": 1.7879417879417881e-06,
      "loss": 0.1205,
      "step": 4037
    },
    {
      "epoch": 4.197505197505198,
      "grad_norm": 1.546683669090271,
      "learning_rate": 1.7856317856317859e-06,
      "loss": 0.0357,
      "step": 4038
    },
    {
      "epoch": 4.198544698544699,
      "grad_norm": 0.24022161960601807,
      "learning_rate": 1.7833217833217834e-06,
      "loss": 0.005,
      "step": 4039
    },
    {
      "epoch": 4.1995841995842,
      "grad_norm": 4.08000373840332,
      "learning_rate": 1.781011781011781e-06,
      "loss": 0.0946,
      "step": 4040
    },
    {
      "epoch": 4.200623700623701,
      "grad_norm": 0.7225533127784729,
      "learning_rate": 1.7787017787017788e-06,
      "loss": 0.0199,
      "step": 4041
    },
    {
      "epoch": 4.201663201663202,
      "grad_norm": 4.201906204223633,
      "learning_rate": 1.7763917763917765e-06,
      "loss": 0.1265,
      "step": 4042
    },
    {
      "epoch": 4.202702702702703,
      "grad_norm": 2.111260175704956,
      "learning_rate": 1.7740817740817743e-06,
      "loss": 0.0775,
      "step": 4043
    },
    {
      "epoch": 4.203742203742204,
      "grad_norm": 3.998500347137451,
      "learning_rate": 1.7717717717717719e-06,
      "loss": 0.3013,
      "step": 4044
    },
    {
      "epoch": 4.204781704781705,
      "grad_norm": 10.227681159973145,
      "learning_rate": 1.7694617694617696e-06,
      "loss": 0.6536,
      "step": 4045
    },
    {
      "epoch": 4.205821205821206,
      "grad_norm": 0.6396321058273315,
      "learning_rate": 1.7671517671517674e-06,
      "loss": 0.0144,
      "step": 4046
    },
    {
      "epoch": 4.206860706860707,
      "grad_norm": 3.9572525024414062,
      "learning_rate": 1.7648417648417652e-06,
      "loss": 0.2184,
      "step": 4047
    },
    {
      "epoch": 4.207900207900208,
      "grad_norm": 4.4264116287231445,
      "learning_rate": 1.7625317625317625e-06,
      "loss": 0.2336,
      "step": 4048
    },
    {
      "epoch": 4.208939708939709,
      "grad_norm": 0.20853319764137268,
      "learning_rate": 1.7602217602217603e-06,
      "loss": 0.004,
      "step": 4049
    },
    {
      "epoch": 4.20997920997921,
      "grad_norm": 1.5663752555847168,
      "learning_rate": 1.757911757911758e-06,
      "loss": 0.0281,
      "step": 4050
    },
    {
      "epoch": 4.211018711018711,
      "grad_norm": 0.04008062928915024,
      "learning_rate": 1.7556017556017558e-06,
      "loss": 0.0012,
      "step": 4051
    },
    {
      "epoch": 4.212058212058212,
      "grad_norm": 5.784019470214844,
      "learning_rate": 1.7532917532917534e-06,
      "loss": 0.2711,
      "step": 4052
    },
    {
      "epoch": 4.213097713097713,
      "grad_norm": 1.9803050756454468,
      "learning_rate": 1.7509817509817511e-06,
      "loss": 0.3233,
      "step": 4053
    },
    {
      "epoch": 4.214137214137214,
      "grad_norm": 2.890364408493042,
      "learning_rate": 1.7486717486717489e-06,
      "loss": 0.052,
      "step": 4054
    },
    {
      "epoch": 4.215176715176715,
      "grad_norm": 0.04758216068148613,
      "learning_rate": 1.7463617463617467e-06,
      "loss": 0.0013,
      "step": 4055
    },
    {
      "epoch": 4.216216216216216,
      "grad_norm": 0.18161475658416748,
      "learning_rate": 1.7440517440517444e-06,
      "loss": 0.0037,
      "step": 4056
    },
    {
      "epoch": 4.217255717255718,
      "grad_norm": 9.276385307312012,
      "learning_rate": 1.7417417417417418e-06,
      "loss": 0.9505,
      "step": 4057
    },
    {
      "epoch": 4.218295218295219,
      "grad_norm": 4.724182605743408,
      "learning_rate": 1.7394317394317395e-06,
      "loss": 0.1682,
      "step": 4058
    },
    {
      "epoch": 4.21933471933472,
      "grad_norm": 0.2898210883140564,
      "learning_rate": 1.7371217371217373e-06,
      "loss": 0.0085,
      "step": 4059
    },
    {
      "epoch": 4.220374220374221,
      "grad_norm": 4.5421833992004395,
      "learning_rate": 1.7348117348117349e-06,
      "loss": 0.4659,
      "step": 4060
    },
    {
      "epoch": 4.221413721413722,
      "grad_norm": 4.829350471496582,
      "learning_rate": 1.7325017325017326e-06,
      "loss": 0.409,
      "step": 4061
    },
    {
      "epoch": 4.222453222453223,
      "grad_norm": 15.786012649536133,
      "learning_rate": 1.7301917301917304e-06,
      "loss": 0.7211,
      "step": 4062
    },
    {
      "epoch": 4.223492723492724,
      "grad_norm": 0.1034480556845665,
      "learning_rate": 1.7278817278817282e-06,
      "loss": 0.0027,
      "step": 4063
    },
    {
      "epoch": 4.224532224532225,
      "grad_norm": 0.5985593795776367,
      "learning_rate": 1.7255717255717257e-06,
      "loss": 0.0105,
      "step": 4064
    },
    {
      "epoch": 4.225571725571726,
      "grad_norm": 3.111302375793457,
      "learning_rate": 1.7232617232617235e-06,
      "loss": 0.0978,
      "step": 4065
    },
    {
      "epoch": 4.226611226611227,
      "grad_norm": 5.8687872886657715,
      "learning_rate": 1.720951720951721e-06,
      "loss": 0.1263,
      "step": 4066
    },
    {
      "epoch": 4.227650727650728,
      "grad_norm": 0.045550454407930374,
      "learning_rate": 1.7186417186417186e-06,
      "loss": 0.0011,
      "step": 4067
    },
    {
      "epoch": 4.228690228690229,
      "grad_norm": 0.04059290140867233,
      "learning_rate": 1.7163317163317164e-06,
      "loss": 0.0006,
      "step": 4068
    },
    {
      "epoch": 4.22972972972973,
      "grad_norm": 2.0094802379608154,
      "learning_rate": 1.7140217140217141e-06,
      "loss": 0.0292,
      "step": 4069
    },
    {
      "epoch": 4.230769230769231,
      "grad_norm": 0.2362472116947174,
      "learning_rate": 1.711711711711712e-06,
      "loss": 0.0057,
      "step": 4070
    },
    {
      "epoch": 4.231808731808732,
      "grad_norm": 0.0013394836569204926,
      "learning_rate": 1.7094017094017097e-06,
      "loss": 0.0,
      "step": 4071
    },
    {
      "epoch": 4.232848232848233,
      "grad_norm": 7.218457221984863,
      "learning_rate": 1.7070917070917072e-06,
      "loss": 0.2895,
      "step": 4072
    },
    {
      "epoch": 4.233887733887734,
      "grad_norm": 1.518581509590149,
      "learning_rate": 1.704781704781705e-06,
      "loss": 0.0419,
      "step": 4073
    },
    {
      "epoch": 4.234927234927235,
      "grad_norm": 2.0033302307128906,
      "learning_rate": 1.7024717024717028e-06,
      "loss": 0.05,
      "step": 4074
    },
    {
      "epoch": 4.235966735966736,
      "grad_norm": 1.7973511219024658,
      "learning_rate": 1.7001617001617e-06,
      "loss": 0.0701,
      "step": 4075
    },
    {
      "epoch": 4.237006237006237,
      "grad_norm": 13.791241645812988,
      "learning_rate": 1.6978516978516979e-06,
      "loss": 0.3143,
      "step": 4076
    },
    {
      "epoch": 4.238045738045738,
      "grad_norm": 6.425209045410156,
      "learning_rate": 1.6955416955416956e-06,
      "loss": 0.0913,
      "step": 4077
    },
    {
      "epoch": 4.239085239085239,
      "grad_norm": 3.037710428237915,
      "learning_rate": 1.6932316932316934e-06,
      "loss": 0.0527,
      "step": 4078
    },
    {
      "epoch": 4.24012474012474,
      "grad_norm": 0.12332512438297272,
      "learning_rate": 1.6909216909216912e-06,
      "loss": 0.0032,
      "step": 4079
    },
    {
      "epoch": 4.241164241164241,
      "grad_norm": 1.4452285766601562,
      "learning_rate": 1.6886116886116887e-06,
      "loss": 0.0545,
      "step": 4080
    },
    {
      "epoch": 4.242203742203742,
      "grad_norm": 4.934514999389648,
      "learning_rate": 1.6863016863016865e-06,
      "loss": 0.0771,
      "step": 4081
    },
    {
      "epoch": 4.243243243243243,
      "grad_norm": 0.048850350081920624,
      "learning_rate": 1.6839916839916843e-06,
      "loss": 0.0013,
      "step": 4082
    },
    {
      "epoch": 4.244282744282744,
      "grad_norm": 5.821188926696777,
      "learning_rate": 1.681681681681682e-06,
      "loss": 0.3531,
      "step": 4083
    },
    {
      "epoch": 4.245322245322245,
      "grad_norm": 2.806713104248047,
      "learning_rate": 1.6793716793716794e-06,
      "loss": 0.0935,
      "step": 4084
    },
    {
      "epoch": 4.246361746361746,
      "grad_norm": 0.1255660057067871,
      "learning_rate": 1.6770616770616771e-06,
      "loss": 0.0027,
      "step": 4085
    },
    {
      "epoch": 4.247401247401247,
      "grad_norm": 7.8570404052734375,
      "learning_rate": 1.674751674751675e-06,
      "loss": 0.3277,
      "step": 4086
    },
    {
      "epoch": 4.248440748440748,
      "grad_norm": 7.057332515716553,
      "learning_rate": 1.6724416724416725e-06,
      "loss": 0.315,
      "step": 4087
    },
    {
      "epoch": 4.24948024948025,
      "grad_norm": 4.830037593841553,
      "learning_rate": 1.6701316701316702e-06,
      "loss": 0.272,
      "step": 4088
    },
    {
      "epoch": 4.25051975051975,
      "grad_norm": 9.737507820129395,
      "learning_rate": 1.667821667821668e-06,
      "loss": 0.5581,
      "step": 4089
    },
    {
      "epoch": 4.251559251559252,
      "grad_norm": 6.508758068084717,
      "learning_rate": 1.6655116655116658e-06,
      "loss": 0.1772,
      "step": 4090
    },
    {
      "epoch": 4.252598752598753,
      "grad_norm": 6.953061580657959,
      "learning_rate": 1.6632016632016635e-06,
      "loss": 0.1369,
      "step": 4091
    },
    {
      "epoch": 4.253638253638254,
      "grad_norm": 3.32177996635437,
      "learning_rate": 1.660891660891661e-06,
      "loss": 0.1571,
      "step": 4092
    },
    {
      "epoch": 4.254677754677755,
      "grad_norm": 5.463006496429443,
      "learning_rate": 1.6585816585816586e-06,
      "loss": 0.0908,
      "step": 4093
    },
    {
      "epoch": 4.255717255717256,
      "grad_norm": 0.16663727164268494,
      "learning_rate": 1.6562716562716564e-06,
      "loss": 0.0056,
      "step": 4094
    },
    {
      "epoch": 4.256756756756757,
      "grad_norm": 5.759579658508301,
      "learning_rate": 1.653961653961654e-06,
      "loss": 0.1231,
      "step": 4095
    },
    {
      "epoch": 4.257796257796258,
      "grad_norm": 2.817511558532715,
      "learning_rate": 1.6516516516516517e-06,
      "loss": 0.089,
      "step": 4096
    },
    {
      "epoch": 4.258835758835759,
      "grad_norm": 5.166245460510254,
      "learning_rate": 1.6493416493416495e-06,
      "loss": 0.1412,
      "step": 4097
    },
    {
      "epoch": 4.25987525987526,
      "grad_norm": 0.08398804813623428,
      "learning_rate": 1.6470316470316473e-06,
      "loss": 0.0015,
      "step": 4098
    },
    {
      "epoch": 4.260914760914761,
      "grad_norm": 3.944993734359741,
      "learning_rate": 1.6447216447216448e-06,
      "loss": 0.076,
      "step": 4099
    },
    {
      "epoch": 4.261954261954262,
      "grad_norm": 1.1812810897827148,
      "learning_rate": 1.6424116424116426e-06,
      "loss": 0.108,
      "step": 4100
    },
    {
      "epoch": 4.262993762993763,
      "grad_norm": 0.8852187991142273,
      "learning_rate": 1.6401016401016403e-06,
      "loss": 0.0227,
      "step": 4101
    },
    {
      "epoch": 4.264033264033264,
      "grad_norm": 8.84179973602295,
      "learning_rate": 1.637791637791638e-06,
      "loss": 0.5302,
      "step": 4102
    },
    {
      "epoch": 4.265072765072765,
      "grad_norm": 1.2681457996368408,
      "learning_rate": 1.6354816354816355e-06,
      "loss": 0.0299,
      "step": 4103
    },
    {
      "epoch": 4.266112266112266,
      "grad_norm": 0.2751096189022064,
      "learning_rate": 1.6331716331716332e-06,
      "loss": 0.0046,
      "step": 4104
    },
    {
      "epoch": 4.267151767151767,
      "grad_norm": 2.4823291301727295,
      "learning_rate": 1.630861630861631e-06,
      "loss": 0.0709,
      "step": 4105
    },
    {
      "epoch": 4.268191268191268,
      "grad_norm": 6.608827590942383,
      "learning_rate": 1.6285516285516288e-06,
      "loss": 0.5985,
      "step": 4106
    },
    {
      "epoch": 4.269230769230769,
      "grad_norm": 0.49379026889801025,
      "learning_rate": 1.6262416262416263e-06,
      "loss": 0.0086,
      "step": 4107
    },
    {
      "epoch": 4.27027027027027,
      "grad_norm": 0.2650279104709625,
      "learning_rate": 1.623931623931624e-06,
      "loss": 0.004,
      "step": 4108
    },
    {
      "epoch": 4.271309771309771,
      "grad_norm": 0.6954243779182434,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 0.2501,
      "step": 4109
    },
    {
      "epoch": 4.272349272349272,
      "grad_norm": 1.0016794204711914,
      "learning_rate": 1.6193116193116196e-06,
      "loss": 0.0508,
      "step": 4110
    },
    {
      "epoch": 4.273388773388773,
      "grad_norm": 0.9003065228462219,
      "learning_rate": 1.617001617001617e-06,
      "loss": 0.0106,
      "step": 4111
    },
    {
      "epoch": 4.274428274428274,
      "grad_norm": 0.20467722415924072,
      "learning_rate": 1.6146916146916147e-06,
      "loss": 0.0035,
      "step": 4112
    },
    {
      "epoch": 4.275467775467775,
      "grad_norm": 7.248074531555176,
      "learning_rate": 1.6123816123816125e-06,
      "loss": 0.8804,
      "step": 4113
    },
    {
      "epoch": 4.276507276507276,
      "grad_norm": 10.930346488952637,
      "learning_rate": 1.6100716100716103e-06,
      "loss": 0.6996,
      "step": 4114
    },
    {
      "epoch": 4.277546777546777,
      "grad_norm": 0.48625805974006653,
      "learning_rate": 1.6077616077616078e-06,
      "loss": 0.0195,
      "step": 4115
    },
    {
      "epoch": 4.278586278586278,
      "grad_norm": 0.8230060935020447,
      "learning_rate": 1.6054516054516056e-06,
      "loss": 0.0188,
      "step": 4116
    },
    {
      "epoch": 4.279625779625779,
      "grad_norm": 0.6208006143569946,
      "learning_rate": 1.6031416031416034e-06,
      "loss": 0.008,
      "step": 4117
    },
    {
      "epoch": 4.28066528066528,
      "grad_norm": 8.017951965332031,
      "learning_rate": 1.6008316008316011e-06,
      "loss": 0.3204,
      "step": 4118
    },
    {
      "epoch": 4.281704781704781,
      "grad_norm": 0.3794807493686676,
      "learning_rate": 1.5985215985215987e-06,
      "loss": 0.0127,
      "step": 4119
    },
    {
      "epoch": 4.282744282744282,
      "grad_norm": 1.9137473106384277,
      "learning_rate": 1.5962115962115962e-06,
      "loss": 0.04,
      "step": 4120
    },
    {
      "epoch": 4.283783783783784,
      "grad_norm": 2.983814239501953,
      "learning_rate": 1.593901593901594e-06,
      "loss": 0.1822,
      "step": 4121
    },
    {
      "epoch": 4.284823284823285,
      "grad_norm": 0.028326358646154404,
      "learning_rate": 1.5915915915915916e-06,
      "loss": 0.0007,
      "step": 4122
    },
    {
      "epoch": 4.285862785862786,
      "grad_norm": 0.3476923704147339,
      "learning_rate": 1.5892815892815893e-06,
      "loss": 0.005,
      "step": 4123
    },
    {
      "epoch": 4.286902286902287,
      "grad_norm": 5.795024394989014,
      "learning_rate": 1.586971586971587e-06,
      "loss": 0.209,
      "step": 4124
    },
    {
      "epoch": 4.287941787941788,
      "grad_norm": 11.100296974182129,
      "learning_rate": 1.5846615846615849e-06,
      "loss": 0.9185,
      "step": 4125
    },
    {
      "epoch": 4.288981288981289,
      "grad_norm": 0.1736786961555481,
      "learning_rate": 1.5823515823515826e-06,
      "loss": 0.0049,
      "step": 4126
    },
    {
      "epoch": 4.29002079002079,
      "grad_norm": 0.010620493441820145,
      "learning_rate": 1.5800415800415802e-06,
      "loss": 0.0002,
      "step": 4127
    },
    {
      "epoch": 4.291060291060291,
      "grad_norm": 0.0010055977618321776,
      "learning_rate": 1.577731577731578e-06,
      "loss": 0.0,
      "step": 4128
    },
    {
      "epoch": 4.292099792099792,
      "grad_norm": 12.3736572265625,
      "learning_rate": 1.5754215754215755e-06,
      "loss": 1.4063,
      "step": 4129
    },
    {
      "epoch": 4.293139293139293,
      "grad_norm": 0.18099883198738098,
      "learning_rate": 1.573111573111573e-06,
      "loss": 0.0055,
      "step": 4130
    },
    {
      "epoch": 4.294178794178794,
      "grad_norm": 0.0030559615697711706,
      "learning_rate": 1.5708015708015708e-06,
      "loss": 0.0001,
      "step": 4131
    },
    {
      "epoch": 4.295218295218295,
      "grad_norm": 0.39262622594833374,
      "learning_rate": 1.5684915684915686e-06,
      "loss": 0.0121,
      "step": 4132
    },
    {
      "epoch": 4.296257796257796,
      "grad_norm": 9.991240501403809,
      "learning_rate": 1.5661815661815664e-06,
      "loss": 0.5465,
      "step": 4133
    },
    {
      "epoch": 4.297297297297297,
      "grad_norm": 0.22802601754665375,
      "learning_rate": 1.5638715638715641e-06,
      "loss": 0.0045,
      "step": 4134
    },
    {
      "epoch": 4.298336798336798,
      "grad_norm": 5.7973198890686035,
      "learning_rate": 1.5615615615615617e-06,
      "loss": 0.2766,
      "step": 4135
    },
    {
      "epoch": 4.299376299376299,
      "grad_norm": 0.11742614209651947,
      "learning_rate": 1.5592515592515594e-06,
      "loss": 0.0024,
      "step": 4136
    },
    {
      "epoch": 4.3004158004158,
      "grad_norm": 8.053942680358887,
      "learning_rate": 1.5569415569415572e-06,
      "loss": 0.36,
      "step": 4137
    },
    {
      "epoch": 4.301455301455301,
      "grad_norm": 8.541512489318848,
      "learning_rate": 1.5546315546315546e-06,
      "loss": 0.683,
      "step": 4138
    },
    {
      "epoch": 4.302494802494802,
      "grad_norm": 0.5943676829338074,
      "learning_rate": 1.5523215523215523e-06,
      "loss": 0.0168,
      "step": 4139
    },
    {
      "epoch": 4.303534303534303,
      "grad_norm": 1.387974500656128,
      "learning_rate": 1.55001155001155e-06,
      "loss": 0.0199,
      "step": 4140
    },
    {
      "epoch": 4.3045738045738045,
      "grad_norm": 0.07895734906196594,
      "learning_rate": 1.5477015477015479e-06,
      "loss": 0.0018,
      "step": 4141
    },
    {
      "epoch": 4.3056133056133055,
      "grad_norm": 0.5686338543891907,
      "learning_rate": 1.5453915453915454e-06,
      "loss": 0.0096,
      "step": 4142
    },
    {
      "epoch": 4.3066528066528065,
      "grad_norm": 12.089524269104004,
      "learning_rate": 1.5430815430815432e-06,
      "loss": 0.7951,
      "step": 4143
    },
    {
      "epoch": 4.3076923076923075,
      "grad_norm": 6.247905731201172,
      "learning_rate": 1.540771540771541e-06,
      "loss": 0.175,
      "step": 4144
    },
    {
      "epoch": 4.3087318087318085,
      "grad_norm": 1.3312230110168457,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 0.026,
      "step": 4145
    },
    {
      "epoch": 4.3097713097713095,
      "grad_norm": 8.780720710754395,
      "learning_rate": 1.5361515361515365e-06,
      "loss": 1.2595,
      "step": 4146
    },
    {
      "epoch": 4.3108108108108105,
      "grad_norm": 8.90927791595459,
      "learning_rate": 1.5338415338415338e-06,
      "loss": 0.3484,
      "step": 4147
    },
    {
      "epoch": 4.3118503118503115,
      "grad_norm": 5.132906436920166,
      "learning_rate": 1.5315315315315316e-06,
      "loss": 0.246,
      "step": 4148
    },
    {
      "epoch": 4.3128898128898125,
      "grad_norm": 2.474851608276367,
      "learning_rate": 1.5292215292215294e-06,
      "loss": 0.0545,
      "step": 4149
    },
    {
      "epoch": 4.313929313929314,
      "grad_norm": 6.656509876251221,
      "learning_rate": 1.526911526911527e-06,
      "loss": 0.3199,
      "step": 4150
    },
    {
      "epoch": 4.314968814968815,
      "grad_norm": 0.038685161620378494,
      "learning_rate": 1.5246015246015247e-06,
      "loss": 0.0009,
      "step": 4151
    },
    {
      "epoch": 4.3160083160083165,
      "grad_norm": 0.49684134125709534,
      "learning_rate": 1.5222915222915225e-06,
      "loss": 0.0098,
      "step": 4152
    },
    {
      "epoch": 4.317047817047817,
      "grad_norm": 0.23896606266498566,
      "learning_rate": 1.5199815199815202e-06,
      "loss": 0.0067,
      "step": 4153
    },
    {
      "epoch": 4.3180873180873185,
      "grad_norm": 7.358112812042236,
      "learning_rate": 1.5176715176715178e-06,
      "loss": 0.1341,
      "step": 4154
    },
    {
      "epoch": 4.3191268191268195,
      "grad_norm": 4.544765949249268,
      "learning_rate": 1.5153615153615155e-06,
      "loss": 0.1136,
      "step": 4155
    },
    {
      "epoch": 4.3201663201663205,
      "grad_norm": 1.8596322536468506,
      "learning_rate": 1.513051513051513e-06,
      "loss": 0.2945,
      "step": 4156
    },
    {
      "epoch": 4.3212058212058215,
      "grad_norm": 10.27206802368164,
      "learning_rate": 1.5107415107415109e-06,
      "loss": 0.6398,
      "step": 4157
    },
    {
      "epoch": 4.3222453222453225,
      "grad_norm": 8.524518013000488,
      "learning_rate": 1.5084315084315084e-06,
      "loss": 0.5755,
      "step": 4158
    },
    {
      "epoch": 4.3232848232848236,
      "grad_norm": 3.3253977298736572,
      "learning_rate": 1.5061215061215062e-06,
      "loss": 0.0561,
      "step": 4159
    },
    {
      "epoch": 4.324324324324325,
      "grad_norm": 0.016905181109905243,
      "learning_rate": 1.503811503811504e-06,
      "loss": 0.0005,
      "step": 4160
    },
    {
      "epoch": 4.325363825363826,
      "grad_norm": 2.6460142135620117,
      "learning_rate": 1.5015015015015017e-06,
      "loss": 0.0686,
      "step": 4161
    },
    {
      "epoch": 4.326403326403327,
      "grad_norm": 11.561341285705566,
      "learning_rate": 1.4991914991914993e-06,
      "loss": 0.9557,
      "step": 4162
    },
    {
      "epoch": 4.327442827442828,
      "grad_norm": 4.719583988189697,
      "learning_rate": 1.496881496881497e-06,
      "loss": 0.3931,
      "step": 4163
    },
    {
      "epoch": 4.328482328482329,
      "grad_norm": 1.616438388824463,
      "learning_rate": 1.4945714945714948e-06,
      "loss": 0.0372,
      "step": 4164
    },
    {
      "epoch": 4.32952182952183,
      "grad_norm": 8.359436988830566,
      "learning_rate": 1.4922614922614922e-06,
      "loss": 0.1088,
      "step": 4165
    },
    {
      "epoch": 4.330561330561331,
      "grad_norm": 8.227208137512207,
      "learning_rate": 1.48995148995149e-06,
      "loss": 0.5338,
      "step": 4166
    },
    {
      "epoch": 4.331600831600832,
      "grad_norm": 1.4397103786468506,
      "learning_rate": 1.4876414876414877e-06,
      "loss": 0.0201,
      "step": 4167
    },
    {
      "epoch": 4.332640332640333,
      "grad_norm": 11.132816314697266,
      "learning_rate": 1.4853314853314855e-06,
      "loss": 0.3741,
      "step": 4168
    },
    {
      "epoch": 4.333679833679834,
      "grad_norm": 5.320253849029541,
      "learning_rate": 1.4830214830214832e-06,
      "loss": 0.2245,
      "step": 4169
    },
    {
      "epoch": 4.334719334719335,
      "grad_norm": 6.323849678039551,
      "learning_rate": 1.4807114807114808e-06,
      "loss": 0.287,
      "step": 4170
    },
    {
      "epoch": 4.335758835758836,
      "grad_norm": 6.769553184509277,
      "learning_rate": 1.4784014784014785e-06,
      "loss": 0.5321,
      "step": 4171
    },
    {
      "epoch": 4.336798336798337,
      "grad_norm": 0.191838338971138,
      "learning_rate": 1.4760914760914763e-06,
      "loss": 0.0044,
      "step": 4172
    },
    {
      "epoch": 4.337837837837838,
      "grad_norm": 7.636790752410889,
      "learning_rate": 1.473781473781474e-06,
      "loss": 0.5625,
      "step": 4173
    },
    {
      "epoch": 4.338877338877339,
      "grad_norm": 2.5185868740081787,
      "learning_rate": 1.4714714714714714e-06,
      "loss": 0.0744,
      "step": 4174
    },
    {
      "epoch": 4.33991683991684,
      "grad_norm": 1.3454338312149048,
      "learning_rate": 1.4691614691614692e-06,
      "loss": 0.0439,
      "step": 4175
    },
    {
      "epoch": 4.340956340956341,
      "grad_norm": 10.914813041687012,
      "learning_rate": 1.466851466851467e-06,
      "loss": 0.5639,
      "step": 4176
    },
    {
      "epoch": 4.341995841995842,
      "grad_norm": 0.06433071941137314,
      "learning_rate": 1.4645414645414645e-06,
      "loss": 0.0016,
      "step": 4177
    },
    {
      "epoch": 4.343035343035343,
      "grad_norm": 10.429954528808594,
      "learning_rate": 1.4622314622314623e-06,
      "loss": 0.484,
      "step": 4178
    },
    {
      "epoch": 4.344074844074844,
      "grad_norm": 7.5033955574035645,
      "learning_rate": 1.45992145992146e-06,
      "loss": 0.13,
      "step": 4179
    },
    {
      "epoch": 4.345114345114345,
      "grad_norm": 11.127370834350586,
      "learning_rate": 1.4576114576114578e-06,
      "loss": 0.1791,
      "step": 4180
    },
    {
      "epoch": 4.346153846153846,
      "grad_norm": 0.08277774602174759,
      "learning_rate": 1.4553014553014556e-06,
      "loss": 0.0026,
      "step": 4181
    },
    {
      "epoch": 4.347193347193347,
      "grad_norm": 11.606966018676758,
      "learning_rate": 1.4529914529914531e-06,
      "loss": 0.7642,
      "step": 4182
    },
    {
      "epoch": 4.348232848232848,
      "grad_norm": 0.2966299057006836,
      "learning_rate": 1.4506814506814507e-06,
      "loss": 0.0065,
      "step": 4183
    },
    {
      "epoch": 4.349272349272349,
      "grad_norm": 0.9831473231315613,
      "learning_rate": 1.4483714483714485e-06,
      "loss": 0.0312,
      "step": 4184
    },
    {
      "epoch": 4.350311850311851,
      "grad_norm": 0.03741445019841194,
      "learning_rate": 1.446061446061446e-06,
      "loss": 0.0009,
      "step": 4185
    },
    {
      "epoch": 4.351351351351352,
      "grad_norm": 1.2937564849853516,
      "learning_rate": 1.4437514437514438e-06,
      "loss": 0.2391,
      "step": 4186
    },
    {
      "epoch": 4.352390852390853,
      "grad_norm": 0.521981954574585,
      "learning_rate": 1.4414414414414416e-06,
      "loss": 0.0237,
      "step": 4187
    },
    {
      "epoch": 4.353430353430354,
      "grad_norm": 0.35736650228500366,
      "learning_rate": 1.4391314391314393e-06,
      "loss": 0.0087,
      "step": 4188
    },
    {
      "epoch": 4.354469854469855,
      "grad_norm": 1.3675105571746826,
      "learning_rate": 1.436821436821437e-06,
      "loss": 0.0442,
      "step": 4189
    },
    {
      "epoch": 4.355509355509356,
      "grad_norm": 4.082332134246826,
      "learning_rate": 1.4345114345114346e-06,
      "loss": 0.3379,
      "step": 4190
    },
    {
      "epoch": 4.356548856548857,
      "grad_norm": 3.070690870285034,
      "learning_rate": 1.4322014322014324e-06,
      "loss": 0.349,
      "step": 4191
    },
    {
      "epoch": 4.357588357588358,
      "grad_norm": 0.032250870019197464,
      "learning_rate": 1.42989142989143e-06,
      "loss": 0.0008,
      "step": 4192
    },
    {
      "epoch": 4.358627858627859,
      "grad_norm": 7.135365962982178,
      "learning_rate": 1.4275814275814275e-06,
      "loss": 0.3598,
      "step": 4193
    },
    {
      "epoch": 4.35966735966736,
      "grad_norm": 0.18401283025741577,
      "learning_rate": 1.4252714252714253e-06,
      "loss": 0.0034,
      "step": 4194
    },
    {
      "epoch": 4.360706860706861,
      "grad_norm": 1.852621078491211,
      "learning_rate": 1.422961422961423e-06,
      "loss": 0.0596,
      "step": 4195
    },
    {
      "epoch": 4.361746361746362,
      "grad_norm": 2.5233609676361084,
      "learning_rate": 1.4206514206514208e-06,
      "loss": 0.0822,
      "step": 4196
    },
    {
      "epoch": 4.362785862785863,
      "grad_norm": 16.242647171020508,
      "learning_rate": 1.4183414183414184e-06,
      "loss": 2.2336,
      "step": 4197
    },
    {
      "epoch": 4.363825363825364,
      "grad_norm": 2.0675549507141113,
      "learning_rate": 1.4160314160314161e-06,
      "loss": 0.0724,
      "step": 4198
    },
    {
      "epoch": 4.364864864864865,
      "grad_norm": 1.6818451881408691,
      "learning_rate": 1.413721413721414e-06,
      "loss": 0.0555,
      "step": 4199
    },
    {
      "epoch": 4.365904365904366,
      "grad_norm": 1.094651699066162,
      "learning_rate": 1.4114114114114117e-06,
      "loss": 0.0243,
      "step": 4200
    },
    {
      "epoch": 4.366943866943867,
      "grad_norm": 0.190381720662117,
      "learning_rate": 1.409101409101409e-06,
      "loss": 0.0055,
      "step": 4201
    },
    {
      "epoch": 4.367983367983368,
      "grad_norm": 4.859400272369385,
      "learning_rate": 1.4067914067914068e-06,
      "loss": 0.1467,
      "step": 4202
    },
    {
      "epoch": 4.369022869022869,
      "grad_norm": 1.171924352645874,
      "learning_rate": 1.4044814044814046e-06,
      "loss": 0.013,
      "step": 4203
    },
    {
      "epoch": 4.37006237006237,
      "grad_norm": 1.6685501337051392,
      "learning_rate": 1.4021714021714023e-06,
      "loss": 0.0578,
      "step": 4204
    },
    {
      "epoch": 4.371101871101871,
      "grad_norm": 10.510454177856445,
      "learning_rate": 1.3998613998613999e-06,
      "loss": 0.3318,
      "step": 4205
    },
    {
      "epoch": 4.372141372141372,
      "grad_norm": 1.731648325920105,
      "learning_rate": 1.3975513975513976e-06,
      "loss": 0.0509,
      "step": 4206
    },
    {
      "epoch": 4.373180873180873,
      "grad_norm": 3.0722243785858154,
      "learning_rate": 1.3952413952413954e-06,
      "loss": 0.0566,
      "step": 4207
    },
    {
      "epoch": 4.374220374220374,
      "grad_norm": 5.660260200500488,
      "learning_rate": 1.3929313929313932e-06,
      "loss": 0.2306,
      "step": 4208
    },
    {
      "epoch": 4.375259875259875,
      "grad_norm": 4.561119079589844,
      "learning_rate": 1.390621390621391e-06,
      "loss": 0.1269,
      "step": 4209
    },
    {
      "epoch": 4.376299376299376,
      "grad_norm": 0.12571902573108673,
      "learning_rate": 1.3883113883113885e-06,
      "loss": 0.0033,
      "step": 4210
    },
    {
      "epoch": 4.377338877338877,
      "grad_norm": 11.195829391479492,
      "learning_rate": 1.386001386001386e-06,
      "loss": 0.6889,
      "step": 4211
    },
    {
      "epoch": 4.378378378378378,
      "grad_norm": 3.5008299350738525,
      "learning_rate": 1.3836913836913838e-06,
      "loss": 0.1269,
      "step": 4212
    },
    {
      "epoch": 4.379417879417879,
      "grad_norm": 0.18679237365722656,
      "learning_rate": 1.3813813813813814e-06,
      "loss": 0.005,
      "step": 4213
    },
    {
      "epoch": 4.38045738045738,
      "grad_norm": 0.2088523656129837,
      "learning_rate": 1.3790713790713791e-06,
      "loss": 0.0034,
      "step": 4214
    },
    {
      "epoch": 4.381496881496881,
      "grad_norm": 0.37627220153808594,
      "learning_rate": 1.376761376761377e-06,
      "loss": 0.0109,
      "step": 4215
    },
    {
      "epoch": 4.382536382536383,
      "grad_norm": 0.010968077927827835,
      "learning_rate": 1.3744513744513747e-06,
      "loss": 0.0002,
      "step": 4216
    },
    {
      "epoch": 4.383575883575883,
      "grad_norm": 1.4702229499816895,
      "learning_rate": 1.3721413721413722e-06,
      "loss": 0.0225,
      "step": 4217
    },
    {
      "epoch": 4.384615384615385,
      "grad_norm": 0.6452034711837769,
      "learning_rate": 1.36983136983137e-06,
      "loss": 0.0138,
      "step": 4218
    },
    {
      "epoch": 4.385654885654886,
      "grad_norm": 1.4516841173171997,
      "learning_rate": 1.3675213675213678e-06,
      "loss": 0.3052,
      "step": 4219
    },
    {
      "epoch": 4.386694386694387,
      "grad_norm": 1.7895114421844482,
      "learning_rate": 1.3652113652113651e-06,
      "loss": 0.0439,
      "step": 4220
    },
    {
      "epoch": 4.387733887733888,
      "grad_norm": 0.15847811102867126,
      "learning_rate": 1.3629013629013629e-06,
      "loss": 0.0031,
      "step": 4221
    },
    {
      "epoch": 4.388773388773389,
      "grad_norm": 6.9709649085998535,
      "learning_rate": 1.3605913605913607e-06,
      "loss": 0.3052,
      "step": 4222
    },
    {
      "epoch": 4.38981288981289,
      "grad_norm": 1.299400806427002,
      "learning_rate": 1.3582813582813584e-06,
      "loss": 0.0293,
      "step": 4223
    },
    {
      "epoch": 4.390852390852391,
      "grad_norm": 11.985542297363281,
      "learning_rate": 1.3559713559713562e-06,
      "loss": 1.561,
      "step": 4224
    },
    {
      "epoch": 4.391891891891892,
      "grad_norm": 14.085444450378418,
      "learning_rate": 1.3536613536613537e-06,
      "loss": 0.8309,
      "step": 4225
    },
    {
      "epoch": 4.392931392931393,
      "grad_norm": 0.11452407389879227,
      "learning_rate": 1.3513513513513515e-06,
      "loss": 0.0021,
      "step": 4226
    },
    {
      "epoch": 4.393970893970894,
      "grad_norm": 3.1590518951416016,
      "learning_rate": 1.3490413490413493e-06,
      "loss": 0.1181,
      "step": 4227
    },
    {
      "epoch": 4.395010395010395,
      "grad_norm": 0.24588535726070404,
      "learning_rate": 1.346731346731347e-06,
      "loss": 0.0038,
      "step": 4228
    },
    {
      "epoch": 4.396049896049896,
      "grad_norm": 0.04706381633877754,
      "learning_rate": 1.3444213444213444e-06,
      "loss": 0.0013,
      "step": 4229
    },
    {
      "epoch": 4.397089397089397,
      "grad_norm": 0.4515562653541565,
      "learning_rate": 1.3421113421113422e-06,
      "loss": 0.0066,
      "step": 4230
    },
    {
      "epoch": 4.398128898128898,
      "grad_norm": 0.4878906011581421,
      "learning_rate": 1.33980133980134e-06,
      "loss": 0.0137,
      "step": 4231
    },
    {
      "epoch": 4.399168399168399,
      "grad_norm": 0.006022973917424679,
      "learning_rate": 1.3374913374913377e-06,
      "loss": 0.0001,
      "step": 4232
    },
    {
      "epoch": 4.4002079002079,
      "grad_norm": 1.3622678518295288,
      "learning_rate": 1.3351813351813352e-06,
      "loss": 0.0421,
      "step": 4233
    },
    {
      "epoch": 4.401247401247401,
      "grad_norm": 5.8447184562683105,
      "learning_rate": 1.332871332871333e-06,
      "loss": 0.1775,
      "step": 4234
    },
    {
      "epoch": 4.402286902286902,
      "grad_norm": 8.86520767211914,
      "learning_rate": 1.3305613305613308e-06,
      "loss": 0.8891,
      "step": 4235
    },
    {
      "epoch": 4.403326403326403,
      "grad_norm": 3.6811683177948,
      "learning_rate": 1.3282513282513285e-06,
      "loss": 0.103,
      "step": 4236
    },
    {
      "epoch": 4.404365904365904,
      "grad_norm": 0.169471874833107,
      "learning_rate": 1.325941325941326e-06,
      "loss": 0.0038,
      "step": 4237
    },
    {
      "epoch": 4.405405405405405,
      "grad_norm": 0.944815456867218,
      "learning_rate": 1.3236313236313237e-06,
      "loss": 0.0175,
      "step": 4238
    },
    {
      "epoch": 4.406444906444906,
      "grad_norm": 0.005099491681903601,
      "learning_rate": 1.3213213213213214e-06,
      "loss": 0.0001,
      "step": 4239
    },
    {
      "epoch": 4.407484407484407,
      "grad_norm": 2.1067795753479004,
      "learning_rate": 1.319011319011319e-06,
      "loss": 0.0942,
      "step": 4240
    },
    {
      "epoch": 4.408523908523908,
      "grad_norm": 5.469954013824463,
      "learning_rate": 1.3167013167013167e-06,
      "loss": 0.126,
      "step": 4241
    },
    {
      "epoch": 4.409563409563409,
      "grad_norm": 0.5456253886222839,
      "learning_rate": 1.3143913143913145e-06,
      "loss": 0.0107,
      "step": 4242
    },
    {
      "epoch": 4.41060291060291,
      "grad_norm": 1.104061484336853,
      "learning_rate": 1.3120813120813123e-06,
      "loss": 0.0221,
      "step": 4243
    },
    {
      "epoch": 4.411642411642411,
      "grad_norm": 0.017622362822294235,
      "learning_rate": 1.30977130977131e-06,
      "loss": 0.0005,
      "step": 4244
    },
    {
      "epoch": 4.412681912681912,
      "grad_norm": 0.05500205606222153,
      "learning_rate": 1.3074613074613076e-06,
      "loss": 0.001,
      "step": 4245
    },
    {
      "epoch": 4.413721413721413,
      "grad_norm": 11.015450477600098,
      "learning_rate": 1.3051513051513054e-06,
      "loss": 1.0906,
      "step": 4246
    },
    {
      "epoch": 4.414760914760914,
      "grad_norm": 12.540595054626465,
      "learning_rate": 1.302841302841303e-06,
      "loss": 0.7457,
      "step": 4247
    },
    {
      "epoch": 4.415800415800415,
      "grad_norm": 0.19104458391666412,
      "learning_rate": 1.3005313005313005e-06,
      "loss": 0.0068,
      "step": 4248
    },
    {
      "epoch": 4.416839916839917,
      "grad_norm": 0.00660668546333909,
      "learning_rate": 1.2982212982212982e-06,
      "loss": 0.0001,
      "step": 4249
    },
    {
      "epoch": 4.417879417879418,
      "grad_norm": 0.6729915142059326,
      "learning_rate": 1.295911295911296e-06,
      "loss": 0.0189,
      "step": 4250
    },
    {
      "epoch": 4.418918918918919,
      "grad_norm": 14.4362154006958,
      "learning_rate": 1.2936012936012938e-06,
      "loss": 0.6613,
      "step": 4251
    },
    {
      "epoch": 4.41995841995842,
      "grad_norm": 2.7787716388702393,
      "learning_rate": 1.2912912912912913e-06,
      "loss": 0.0493,
      "step": 4252
    },
    {
      "epoch": 4.420997920997921,
      "grad_norm": 5.139069557189941,
      "learning_rate": 1.288981288981289e-06,
      "loss": 0.6069,
      "step": 4253
    },
    {
      "epoch": 4.422037422037422,
      "grad_norm": 6.8310465812683105,
      "learning_rate": 1.2866712866712869e-06,
      "loss": 0.1973,
      "step": 4254
    },
    {
      "epoch": 4.423076923076923,
      "grad_norm": 1.6439793109893799,
      "learning_rate": 1.2843612843612846e-06,
      "loss": 0.0401,
      "step": 4255
    },
    {
      "epoch": 4.424116424116424,
      "grad_norm": 5.19002103805542,
      "learning_rate": 1.282051282051282e-06,
      "loss": 0.2427,
      "step": 4256
    },
    {
      "epoch": 4.425155925155925,
      "grad_norm": 6.832848072052002,
      "learning_rate": 1.2797412797412797e-06,
      "loss": 0.4795,
      "step": 4257
    },
    {
      "epoch": 4.426195426195426,
      "grad_norm": 10.052824974060059,
      "learning_rate": 1.2774312774312775e-06,
      "loss": 0.688,
      "step": 4258
    },
    {
      "epoch": 4.427234927234927,
      "grad_norm": 1.8471096754074097,
      "learning_rate": 1.2751212751212753e-06,
      "loss": 0.0632,
      "step": 4259
    },
    {
      "epoch": 4.428274428274428,
      "grad_norm": 2.3802993297576904,
      "learning_rate": 1.2728112728112728e-06,
      "loss": 0.0726,
      "step": 4260
    },
    {
      "epoch": 4.429313929313929,
      "grad_norm": 1.7276345491409302,
      "learning_rate": 1.2705012705012706e-06,
      "loss": 0.0318,
      "step": 4261
    },
    {
      "epoch": 4.43035343035343,
      "grad_norm": 1.354040503501892,
      "learning_rate": 1.2681912681912684e-06,
      "loss": 0.282,
      "step": 4262
    },
    {
      "epoch": 4.4313929313929314,
      "grad_norm": 4.793944358825684,
      "learning_rate": 1.2658812658812661e-06,
      "loss": 0.2987,
      "step": 4263
    },
    {
      "epoch": 4.4324324324324325,
      "grad_norm": 5.113119125366211,
      "learning_rate": 1.263571263571264e-06,
      "loss": 0.1797,
      "step": 4264
    },
    {
      "epoch": 4.4334719334719335,
      "grad_norm": 4.361006259918213,
      "learning_rate": 1.2612612612612613e-06,
      "loss": 0.3196,
      "step": 4265
    },
    {
      "epoch": 4.4345114345114345,
      "grad_norm": 2.3230502605438232,
      "learning_rate": 1.258951258951259e-06,
      "loss": 0.0671,
      "step": 4266
    },
    {
      "epoch": 4.4355509355509355,
      "grad_norm": 8.467252731323242,
      "learning_rate": 1.2566412566412568e-06,
      "loss": 0.2399,
      "step": 4267
    },
    {
      "epoch": 4.4365904365904365,
      "grad_norm": 0.09298921376466751,
      "learning_rate": 1.2543312543312543e-06,
      "loss": 0.0017,
      "step": 4268
    },
    {
      "epoch": 4.4376299376299375,
      "grad_norm": 4.662631034851074,
      "learning_rate": 1.2520212520212521e-06,
      "loss": 0.2203,
      "step": 4269
    },
    {
      "epoch": 4.4386694386694385,
      "grad_norm": 7.328342914581299,
      "learning_rate": 1.2497112497112499e-06,
      "loss": 0.2689,
      "step": 4270
    },
    {
      "epoch": 4.4397089397089395,
      "grad_norm": 0.9401403665542603,
      "learning_rate": 1.2474012474012476e-06,
      "loss": 0.032,
      "step": 4271
    },
    {
      "epoch": 4.4407484407484406,
      "grad_norm": 6.420727729797363,
      "learning_rate": 1.2450912450912452e-06,
      "loss": 0.3668,
      "step": 4272
    },
    {
      "epoch": 4.441787941787942,
      "grad_norm": 3.3101179599761963,
      "learning_rate": 1.242781242781243e-06,
      "loss": 0.131,
      "step": 4273
    },
    {
      "epoch": 4.442827442827443,
      "grad_norm": 0.010817071422934532,
      "learning_rate": 1.2404712404712405e-06,
      "loss": 0.0002,
      "step": 4274
    },
    {
      "epoch": 4.443866943866944,
      "grad_norm": 6.34616756439209,
      "learning_rate": 1.2381612381612383e-06,
      "loss": 0.0606,
      "step": 4275
    },
    {
      "epoch": 4.444906444906445,
      "grad_norm": 4.0047712326049805,
      "learning_rate": 1.2358512358512358e-06,
      "loss": 0.4552,
      "step": 4276
    },
    {
      "epoch": 4.445945945945946,
      "grad_norm": 7.248794078826904,
      "learning_rate": 1.2335412335412336e-06,
      "loss": 0.2012,
      "step": 4277
    },
    {
      "epoch": 4.446985446985447,
      "grad_norm": 4.6254425048828125,
      "learning_rate": 1.2312312312312314e-06,
      "loss": 0.3375,
      "step": 4278
    },
    {
      "epoch": 4.448024948024948,
      "grad_norm": 4.371951103210449,
      "learning_rate": 1.2289212289212291e-06,
      "loss": 0.0718,
      "step": 4279
    },
    {
      "epoch": 4.4490644490644495,
      "grad_norm": 0.24949759244918823,
      "learning_rate": 1.2266112266112267e-06,
      "loss": 0.0044,
      "step": 4280
    },
    {
      "epoch": 4.45010395010395,
      "grad_norm": 0.3559669256210327,
      "learning_rate": 1.2243012243012243e-06,
      "loss": 0.004,
      "step": 4281
    },
    {
      "epoch": 4.451143451143452,
      "grad_norm": 3.888190984725952,
      "learning_rate": 1.221991221991222e-06,
      "loss": 0.04,
      "step": 4282
    },
    {
      "epoch": 4.452182952182953,
      "grad_norm": 6.422785758972168,
      "learning_rate": 1.2196812196812198e-06,
      "loss": 0.2481,
      "step": 4283
    },
    {
      "epoch": 4.453222453222454,
      "grad_norm": 1.392903447151184,
      "learning_rate": 1.2173712173712176e-06,
      "loss": 0.021,
      "step": 4284
    },
    {
      "epoch": 4.454261954261955,
      "grad_norm": 0.9166313409805298,
      "learning_rate": 1.2150612150612151e-06,
      "loss": 0.0304,
      "step": 4285
    },
    {
      "epoch": 4.455301455301456,
      "grad_norm": 0.4601767957210541,
      "learning_rate": 1.2127512127512129e-06,
      "loss": 0.0083,
      "step": 4286
    },
    {
      "epoch": 4.456340956340957,
      "grad_norm": 7.392816543579102,
      "learning_rate": 1.2104412104412106e-06,
      "loss": 0.3828,
      "step": 4287
    },
    {
      "epoch": 4.457380457380458,
      "grad_norm": 0.008619803935289383,
      "learning_rate": 1.2081312081312082e-06,
      "loss": 0.0002,
      "step": 4288
    },
    {
      "epoch": 4.458419958419959,
      "grad_norm": 0.06648194044828415,
      "learning_rate": 1.205821205821206e-06,
      "loss": 0.0024,
      "step": 4289
    },
    {
      "epoch": 4.45945945945946,
      "grad_norm": 0.06525490432977676,
      "learning_rate": 1.2035112035112035e-06,
      "loss": 0.0012,
      "step": 4290
    },
    {
      "epoch": 4.460498960498961,
      "grad_norm": 7.3383469581604,
      "learning_rate": 1.2012012012012013e-06,
      "loss": 0.4375,
      "step": 4291
    },
    {
      "epoch": 4.461538461538462,
      "grad_norm": 2.6314022541046143,
      "learning_rate": 1.198891198891199e-06,
      "loss": 0.0536,
      "step": 4292
    },
    {
      "epoch": 4.462577962577963,
      "grad_norm": 5.249322891235352,
      "learning_rate": 1.1965811965811968e-06,
      "loss": 0.1795,
      "step": 4293
    },
    {
      "epoch": 4.463617463617464,
      "grad_norm": 7.783726692199707,
      "learning_rate": 1.1942711942711944e-06,
      "loss": 0.5374,
      "step": 4294
    },
    {
      "epoch": 4.464656964656965,
      "grad_norm": 9.920394897460938,
      "learning_rate": 1.191961191961192e-06,
      "loss": 1.2068,
      "step": 4295
    },
    {
      "epoch": 4.465696465696466,
      "grad_norm": 0.939612627029419,
      "learning_rate": 1.1896511896511897e-06,
      "loss": 0.0164,
      "step": 4296
    },
    {
      "epoch": 4.466735966735967,
      "grad_norm": 0.01430198922753334,
      "learning_rate": 1.1873411873411875e-06,
      "loss": 0.0003,
      "step": 4297
    },
    {
      "epoch": 4.467775467775468,
      "grad_norm": 6.761784076690674,
      "learning_rate": 1.1850311850311852e-06,
      "loss": 0.298,
      "step": 4298
    },
    {
      "epoch": 4.468814968814969,
      "grad_norm": 0.29458633065223694,
      "learning_rate": 1.1827211827211828e-06,
      "loss": 0.0058,
      "step": 4299
    },
    {
      "epoch": 4.46985446985447,
      "grad_norm": 5.258695125579834,
      "learning_rate": 1.1804111804111806e-06,
      "loss": 0.0941,
      "step": 4300
    },
    {
      "epoch": 4.470893970893971,
      "grad_norm": 6.842070579528809,
      "learning_rate": 1.1781011781011781e-06,
      "loss": 0.3605,
      "step": 4301
    },
    {
      "epoch": 4.471933471933472,
      "grad_norm": 3.9812428951263428,
      "learning_rate": 1.1757911757911759e-06,
      "loss": 0.268,
      "step": 4302
    },
    {
      "epoch": 4.472972972972973,
      "grad_norm": 5.772032737731934,
      "learning_rate": 1.1734811734811734e-06,
      "loss": 0.3061,
      "step": 4303
    },
    {
      "epoch": 4.474012474012474,
      "grad_norm": 0.419314444065094,
      "learning_rate": 1.1711711711711712e-06,
      "loss": 0.0084,
      "step": 4304
    },
    {
      "epoch": 4.475051975051975,
      "grad_norm": 0.5556364059448242,
      "learning_rate": 1.168861168861169e-06,
      "loss": 0.0126,
      "step": 4305
    },
    {
      "epoch": 4.476091476091476,
      "grad_norm": 8.333703994750977,
      "learning_rate": 1.1665511665511667e-06,
      "loss": 1.235,
      "step": 4306
    },
    {
      "epoch": 4.477130977130977,
      "grad_norm": 1.7269006967544556,
      "learning_rate": 1.1642411642411643e-06,
      "loss": 0.0509,
      "step": 4307
    },
    {
      "epoch": 4.478170478170478,
      "grad_norm": 5.627634048461914,
      "learning_rate": 1.161931161931162e-06,
      "loss": 0.2021,
      "step": 4308
    },
    {
      "epoch": 4.479209979209979,
      "grad_norm": 0.252849280834198,
      "learning_rate": 1.1596211596211596e-06,
      "loss": 0.0046,
      "step": 4309
    },
    {
      "epoch": 4.48024948024948,
      "grad_norm": 0.008383556269109249,
      "learning_rate": 1.1573111573111574e-06,
      "loss": 0.0002,
      "step": 4310
    },
    {
      "epoch": 4.481288981288981,
      "grad_norm": 0.9866765141487122,
      "learning_rate": 1.1550011550011552e-06,
      "loss": 0.018,
      "step": 4311
    },
    {
      "epoch": 4.482328482328482,
      "grad_norm": 0.01083228550851345,
      "learning_rate": 1.1526911526911527e-06,
      "loss": 0.0003,
      "step": 4312
    },
    {
      "epoch": 4.483367983367984,
      "grad_norm": 0.04575052112340927,
      "learning_rate": 1.1503811503811505e-06,
      "loss": 0.0012,
      "step": 4313
    },
    {
      "epoch": 4.484407484407485,
      "grad_norm": 0.0022366337943822145,
      "learning_rate": 1.1480711480711482e-06,
      "loss": 0.0,
      "step": 4314
    },
    {
      "epoch": 4.485446985446986,
      "grad_norm": 8.27820873260498,
      "learning_rate": 1.1457611457611458e-06,
      "loss": 0.3146,
      "step": 4315
    },
    {
      "epoch": 4.486486486486487,
      "grad_norm": 0.008490752428770065,
      "learning_rate": 1.1434511434511436e-06,
      "loss": 0.0002,
      "step": 4316
    },
    {
      "epoch": 4.487525987525988,
      "grad_norm": 0.03850068524479866,
      "learning_rate": 1.1411411411411411e-06,
      "loss": 0.0005,
      "step": 4317
    },
    {
      "epoch": 4.488565488565489,
      "grad_norm": 0.014865309000015259,
      "learning_rate": 1.1388311388311389e-06,
      "loss": 0.0005,
      "step": 4318
    },
    {
      "epoch": 4.48960498960499,
      "grad_norm": 8.617644309997559,
      "learning_rate": 1.1365211365211367e-06,
      "loss": 0.2046,
      "step": 4319
    },
    {
      "epoch": 4.490644490644491,
      "grad_norm": 0.02923273667693138,
      "learning_rate": 1.1342111342111344e-06,
      "loss": 0.0004,
      "step": 4320
    },
    {
      "epoch": 4.491683991683992,
      "grad_norm": 2.4688708782196045,
      "learning_rate": 1.131901131901132e-06,
      "loss": 0.0451,
      "step": 4321
    },
    {
      "epoch": 4.492723492723493,
      "grad_norm": 4.8640546798706055,
      "learning_rate": 1.1295911295911297e-06,
      "loss": 0.1706,
      "step": 4322
    },
    {
      "epoch": 4.493762993762994,
      "grad_norm": 0.36775505542755127,
      "learning_rate": 1.1272811272811273e-06,
      "loss": 0.0089,
      "step": 4323
    },
    {
      "epoch": 4.494802494802495,
      "grad_norm": 11.362357139587402,
      "learning_rate": 1.124971124971125e-06,
      "loss": 0.3145,
      "step": 4324
    },
    {
      "epoch": 4.495841995841996,
      "grad_norm": 10.543671607971191,
      "learning_rate": 1.1226611226611228e-06,
      "loss": 0.6145,
      "step": 4325
    },
    {
      "epoch": 4.496881496881497,
      "grad_norm": 5.450479030609131,
      "learning_rate": 1.1203511203511204e-06,
      "loss": 0.1512,
      "step": 4326
    },
    {
      "epoch": 4.497920997920998,
      "grad_norm": 0.10827375948429108,
      "learning_rate": 1.1180411180411182e-06,
      "loss": 0.0024,
      "step": 4327
    },
    {
      "epoch": 4.498960498960499,
      "grad_norm": 8.659412384033203,
      "learning_rate": 1.115731115731116e-06,
      "loss": 0.9051,
      "step": 4328
    },
    {
      "epoch": 4.5,
      "grad_norm": 10.345586776733398,
      "learning_rate": 1.1134211134211135e-06,
      "loss": 1.4444,
      "step": 4329
    },
    {
      "epoch": 4.501039501039501,
      "grad_norm": 1.9703149795532227,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.0309,
      "step": 4330
    },
    {
      "epoch": 4.502079002079002,
      "grad_norm": 4.598181247711182,
      "learning_rate": 1.1088011088011088e-06,
      "loss": 0.2541,
      "step": 4331
    },
    {
      "epoch": 4.503118503118503,
      "grad_norm": 7.416522026062012,
      "learning_rate": 1.1064911064911066e-06,
      "loss": 0.2692,
      "step": 4332
    },
    {
      "epoch": 4.504158004158004,
      "grad_norm": 7.817275524139404,
      "learning_rate": 1.1041811041811043e-06,
      "loss": 0.6485,
      "step": 4333
    },
    {
      "epoch": 4.505197505197505,
      "grad_norm": 5.300344944000244,
      "learning_rate": 1.1018711018711021e-06,
      "loss": 0.4363,
      "step": 4334
    },
    {
      "epoch": 4.506237006237006,
      "grad_norm": 10.25655460357666,
      "learning_rate": 1.0995610995610997e-06,
      "loss": 0.5223,
      "step": 4335
    },
    {
      "epoch": 4.507276507276507,
      "grad_norm": 5.892490863800049,
      "learning_rate": 1.0972510972510974e-06,
      "loss": 0.3733,
      "step": 4336
    },
    {
      "epoch": 4.508316008316008,
      "grad_norm": 8.785927772521973,
      "learning_rate": 1.094941094941095e-06,
      "loss": 0.6725,
      "step": 4337
    },
    {
      "epoch": 4.509355509355509,
      "grad_norm": 0.9067697525024414,
      "learning_rate": 1.0926310926310928e-06,
      "loss": 0.0146,
      "step": 4338
    },
    {
      "epoch": 4.51039501039501,
      "grad_norm": 5.748349666595459,
      "learning_rate": 1.0903210903210903e-06,
      "loss": 0.636,
      "step": 4339
    },
    {
      "epoch": 4.511434511434511,
      "grad_norm": 0.0022521372884511948,
      "learning_rate": 1.088011088011088e-06,
      "loss": 0.0001,
      "step": 4340
    },
    {
      "epoch": 4.512474012474012,
      "grad_norm": 0.20115229487419128,
      "learning_rate": 1.0857010857010858e-06,
      "loss": 0.0045,
      "step": 4341
    },
    {
      "epoch": 4.513513513513513,
      "grad_norm": 4.627670764923096,
      "learning_rate": 1.0833910833910836e-06,
      "loss": 0.1453,
      "step": 4342
    },
    {
      "epoch": 4.514553014553014,
      "grad_norm": 4.924317359924316,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 0.1042,
      "step": 4343
    },
    {
      "epoch": 4.515592515592516,
      "grad_norm": 2.5569543838500977,
      "learning_rate": 1.0787710787710787e-06,
      "loss": 0.0342,
      "step": 4344
    },
    {
      "epoch": 4.516632016632016,
      "grad_norm": 6.637704372406006,
      "learning_rate": 1.0764610764610765e-06,
      "loss": 0.6174,
      "step": 4345
    },
    {
      "epoch": 4.517671517671518,
      "grad_norm": 0.4843396246433258,
      "learning_rate": 1.0741510741510743e-06,
      "loss": 0.014,
      "step": 4346
    },
    {
      "epoch": 4.518711018711018,
      "grad_norm": 5.531055927276611,
      "learning_rate": 1.071841071841072e-06,
      "loss": 0.2442,
      "step": 4347
    },
    {
      "epoch": 4.51975051975052,
      "grad_norm": 2.8017818927764893,
      "learning_rate": 1.0695310695310696e-06,
      "loss": 0.0746,
      "step": 4348
    },
    {
      "epoch": 4.520790020790021,
      "grad_norm": 7.472479820251465,
      "learning_rate": 1.0672210672210673e-06,
      "loss": 0.1791,
      "step": 4349
    },
    {
      "epoch": 4.521829521829522,
      "grad_norm": 1.5929454565048218,
      "learning_rate": 1.064911064911065e-06,
      "loss": 0.0339,
      "step": 4350
    },
    {
      "epoch": 4.522869022869023,
      "grad_norm": 2.3107383251190186,
      "learning_rate": 1.0626010626010627e-06,
      "loss": 0.0812,
      "step": 4351
    },
    {
      "epoch": 4.523908523908524,
      "grad_norm": 9.258277893066406,
      "learning_rate": 1.0602910602910604e-06,
      "loss": 0.6373,
      "step": 4352
    },
    {
      "epoch": 4.524948024948025,
      "grad_norm": 20.639976501464844,
      "learning_rate": 1.057981057981058e-06,
      "loss": 0.1888,
      "step": 4353
    },
    {
      "epoch": 4.525987525987526,
      "grad_norm": 0.026088906452059746,
      "learning_rate": 1.0556710556710558e-06,
      "loss": 0.0006,
      "step": 4354
    },
    {
      "epoch": 4.527027027027027,
      "grad_norm": 8.579378128051758,
      "learning_rate": 1.0533610533610535e-06,
      "loss": 0.5171,
      "step": 4355
    },
    {
      "epoch": 4.528066528066528,
      "grad_norm": 5.382491588592529,
      "learning_rate": 1.051051051051051e-06,
      "loss": 0.0538,
      "step": 4356
    },
    {
      "epoch": 4.529106029106029,
      "grad_norm": 8.6734037399292,
      "learning_rate": 1.0487410487410488e-06,
      "loss": 0.9482,
      "step": 4357
    },
    {
      "epoch": 4.53014553014553,
      "grad_norm": 1.555879831314087,
      "learning_rate": 1.0464310464310464e-06,
      "loss": 0.0433,
      "step": 4358
    },
    {
      "epoch": 4.531185031185031,
      "grad_norm": 6.786050796508789,
      "learning_rate": 1.0441210441210442e-06,
      "loss": 0.147,
      "step": 4359
    },
    {
      "epoch": 4.532224532224532,
      "grad_norm": 5.747766017913818,
      "learning_rate": 1.041811041811042e-06,
      "loss": 0.132,
      "step": 4360
    },
    {
      "epoch": 4.533264033264033,
      "grad_norm": 1.9380532503128052,
      "learning_rate": 1.0395010395010397e-06,
      "loss": 0.2936,
      "step": 4361
    },
    {
      "epoch": 4.534303534303534,
      "grad_norm": 9.890674591064453,
      "learning_rate": 1.0371910371910373e-06,
      "loss": 0.1638,
      "step": 4362
    },
    {
      "epoch": 4.535343035343035,
      "grad_norm": 2.7257440090179443,
      "learning_rate": 1.034881034881035e-06,
      "loss": 0.0508,
      "step": 4363
    },
    {
      "epoch": 4.536382536382536,
      "grad_norm": 8.347380638122559,
      "learning_rate": 1.0325710325710326e-06,
      "loss": 0.7168,
      "step": 4364
    },
    {
      "epoch": 4.537422037422037,
      "grad_norm": 6.967242240905762,
      "learning_rate": 1.0302610302610303e-06,
      "loss": 0.3137,
      "step": 4365
    },
    {
      "epoch": 4.538461538461538,
      "grad_norm": 2.8173623085021973,
      "learning_rate": 1.027951027951028e-06,
      "loss": 0.092,
      "step": 4366
    },
    {
      "epoch": 4.539501039501039,
      "grad_norm": 0.10028092563152313,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 0.002,
      "step": 4367
    },
    {
      "epoch": 4.54054054054054,
      "grad_norm": 4.162265300750732,
      "learning_rate": 1.0233310233310234e-06,
      "loss": 0.1445,
      "step": 4368
    },
    {
      "epoch": 4.541580041580041,
      "grad_norm": 0.7075111865997314,
      "learning_rate": 1.0210210210210212e-06,
      "loss": 0.0087,
      "step": 4369
    },
    {
      "epoch": 4.542619542619542,
      "grad_norm": 10.8870210647583,
      "learning_rate": 1.0187110187110188e-06,
      "loss": 1.9218,
      "step": 4370
    },
    {
      "epoch": 4.543659043659043,
      "grad_norm": 0.6903623938560486,
      "learning_rate": 1.0164010164010165e-06,
      "loss": 0.0273,
      "step": 4371
    },
    {
      "epoch": 4.544698544698544,
      "grad_norm": 8.723344802856445,
      "learning_rate": 1.014091014091014e-06,
      "loss": 0.5867,
      "step": 4372
    },
    {
      "epoch": 4.545738045738045,
      "grad_norm": 0.913175106048584,
      "learning_rate": 1.0117810117810119e-06,
      "loss": 0.0585,
      "step": 4373
    },
    {
      "epoch": 4.546777546777546,
      "grad_norm": 11.532629013061523,
      "learning_rate": 1.0094710094710096e-06,
      "loss": 0.5133,
      "step": 4374
    },
    {
      "epoch": 4.547817047817047,
      "grad_norm": 1.2006914615631104,
      "learning_rate": 1.0071610071610072e-06,
      "loss": 0.0428,
      "step": 4375
    },
    {
      "epoch": 4.548856548856548,
      "grad_norm": 2.0305368900299072,
      "learning_rate": 1.004851004851005e-06,
      "loss": 0.0362,
      "step": 4376
    },
    {
      "epoch": 4.54989604989605,
      "grad_norm": 0.8395141959190369,
      "learning_rate": 1.0025410025410027e-06,
      "loss": 0.0093,
      "step": 4377
    },
    {
      "epoch": 4.5509355509355505,
      "grad_norm": 0.9320726990699768,
      "learning_rate": 1.0002310002310003e-06,
      "loss": 0.0164,
      "step": 4378
    },
    {
      "epoch": 4.551975051975052,
      "grad_norm": 9.359164237976074,
      "learning_rate": 9.97920997920998e-07,
      "loss": 0.3419,
      "step": 4379
    },
    {
      "epoch": 4.553014553014553,
      "grad_norm": 7.8454508781433105,
      "learning_rate": 9.956109956109956e-07,
      "loss": 0.444,
      "step": 4380
    },
    {
      "epoch": 4.554054054054054,
      "grad_norm": 0.4756433665752411,
      "learning_rate": 9.933009933009934e-07,
      "loss": 0.012,
      "step": 4381
    },
    {
      "epoch": 4.555093555093555,
      "grad_norm": 0.028461869806051254,
      "learning_rate": 9.909909909909911e-07,
      "loss": 0.0007,
      "step": 4382
    },
    {
      "epoch": 4.556133056133056,
      "grad_norm": 10.92263126373291,
      "learning_rate": 9.886809886809889e-07,
      "loss": 0.7289,
      "step": 4383
    },
    {
      "epoch": 4.557172557172557,
      "grad_norm": 0.09633204340934753,
      "learning_rate": 9.863709863709864e-07,
      "loss": 0.0025,
      "step": 4384
    },
    {
      "epoch": 4.558212058212058,
      "grad_norm": 1.7298146486282349,
      "learning_rate": 9.84060984060984e-07,
      "loss": 0.0215,
      "step": 4385
    },
    {
      "epoch": 4.5592515592515594,
      "grad_norm": 5.163122177124023,
      "learning_rate": 9.817509817509818e-07,
      "loss": 0.2054,
      "step": 4386
    },
    {
      "epoch": 4.5602910602910605,
      "grad_norm": 0.41875556111335754,
      "learning_rate": 9.794409794409795e-07,
      "loss": 0.0072,
      "step": 4387
    },
    {
      "epoch": 4.5613305613305615,
      "grad_norm": 8.408621788024902,
      "learning_rate": 9.771309771309773e-07,
      "loss": 0.2993,
      "step": 4388
    },
    {
      "epoch": 4.5623700623700625,
      "grad_norm": 3.3478004932403564,
      "learning_rate": 9.748209748209749e-07,
      "loss": 0.083,
      "step": 4389
    },
    {
      "epoch": 4.5634095634095635,
      "grad_norm": 9.404569625854492,
      "learning_rate": 9.725109725109726e-07,
      "loss": 0.3557,
      "step": 4390
    },
    {
      "epoch": 4.5644490644490645,
      "grad_norm": 2.7835049629211426,
      "learning_rate": 9.702009702009704e-07,
      "loss": 0.0758,
      "step": 4391
    },
    {
      "epoch": 4.5654885654885655,
      "grad_norm": 7.355697154998779,
      "learning_rate": 9.67890967890968e-07,
      "loss": 0.2352,
      "step": 4392
    },
    {
      "epoch": 4.5665280665280665,
      "grad_norm": 0.16567951440811157,
      "learning_rate": 9.655809655809657e-07,
      "loss": 0.0044,
      "step": 4393
    },
    {
      "epoch": 4.5675675675675675,
      "grad_norm": 0.0037194276228547096,
      "learning_rate": 9.632709632709633e-07,
      "loss": 0.0001,
      "step": 4394
    },
    {
      "epoch": 4.5686070686070686,
      "grad_norm": 0.23683440685272217,
      "learning_rate": 9.60960960960961e-07,
      "loss": 0.0066,
      "step": 4395
    },
    {
      "epoch": 4.56964656964657,
      "grad_norm": 0.004315780010074377,
      "learning_rate": 9.586509586509588e-07,
      "loss": 0.0001,
      "step": 4396
    },
    {
      "epoch": 4.570686070686071,
      "grad_norm": 0.0919046625494957,
      "learning_rate": 9.563409563409566e-07,
      "loss": 0.0021,
      "step": 4397
    },
    {
      "epoch": 4.571725571725572,
      "grad_norm": 0.9975956082344055,
      "learning_rate": 9.540309540309541e-07,
      "loss": 0.0538,
      "step": 4398
    },
    {
      "epoch": 4.572765072765073,
      "grad_norm": 0.11247812211513519,
      "learning_rate": 9.517209517209518e-07,
      "loss": 0.0032,
      "step": 4399
    },
    {
      "epoch": 4.573804573804574,
      "grad_norm": 4.416209697723389,
      "learning_rate": 9.494109494109494e-07,
      "loss": 0.1403,
      "step": 4400
    },
    {
      "epoch": 4.574844074844075,
      "grad_norm": 0.0890134945511818,
      "learning_rate": 9.471009471009472e-07,
      "loss": 0.0014,
      "step": 4401
    },
    {
      "epoch": 4.575883575883576,
      "grad_norm": 0.5805600881576538,
      "learning_rate": 9.44790944790945e-07,
      "loss": 0.0125,
      "step": 4402
    },
    {
      "epoch": 4.576923076923077,
      "grad_norm": 1.9322223663330078,
      "learning_rate": 9.424809424809425e-07,
      "loss": 0.0663,
      "step": 4403
    },
    {
      "epoch": 4.577962577962578,
      "grad_norm": 3.145690679550171,
      "learning_rate": 9.401709401709402e-07,
      "loss": 0.0971,
      "step": 4404
    },
    {
      "epoch": 4.579002079002079,
      "grad_norm": 2.712754011154175,
      "learning_rate": 9.37860937860938e-07,
      "loss": 0.1244,
      "step": 4405
    },
    {
      "epoch": 4.58004158004158,
      "grad_norm": 0.07731432467699051,
      "learning_rate": 9.355509355509356e-07,
      "loss": 0.0019,
      "step": 4406
    },
    {
      "epoch": 4.581081081081081,
      "grad_norm": 0.10027692466974258,
      "learning_rate": 9.332409332409333e-07,
      "loss": 0.0029,
      "step": 4407
    },
    {
      "epoch": 4.582120582120583,
      "grad_norm": 3.366645574569702,
      "learning_rate": 9.30930930930931e-07,
      "loss": 0.1166,
      "step": 4408
    },
    {
      "epoch": 4.583160083160083,
      "grad_norm": 8.035176277160645,
      "learning_rate": 9.286209286209287e-07,
      "loss": 0.2173,
      "step": 4409
    },
    {
      "epoch": 4.584199584199585,
      "grad_norm": 3.704331159591675,
      "learning_rate": 9.263109263109264e-07,
      "loss": 0.1451,
      "step": 4410
    },
    {
      "epoch": 4.585239085239085,
      "grad_norm": 1.452633023262024,
      "learning_rate": 9.240009240009241e-07,
      "loss": 0.0318,
      "step": 4411
    },
    {
      "epoch": 4.586278586278587,
      "grad_norm": 5.019649028778076,
      "learning_rate": 9.216909216909217e-07,
      "loss": 0.2807,
      "step": 4412
    },
    {
      "epoch": 4.587318087318088,
      "grad_norm": 5.823719501495361,
      "learning_rate": 9.193809193809195e-07,
      "loss": 0.1205,
      "step": 4413
    },
    {
      "epoch": 4.588357588357589,
      "grad_norm": 1.7746714353561401,
      "learning_rate": 9.170709170709171e-07,
      "loss": 0.0313,
      "step": 4414
    },
    {
      "epoch": 4.58939708939709,
      "grad_norm": 7.47543478012085,
      "learning_rate": 9.147609147609149e-07,
      "loss": 0.2132,
      "step": 4415
    },
    {
      "epoch": 4.590436590436591,
      "grad_norm": 0.004619097337126732,
      "learning_rate": 9.124509124509125e-07,
      "loss": 0.0001,
      "step": 4416
    },
    {
      "epoch": 4.591476091476092,
      "grad_norm": 2.1546518802642822,
      "learning_rate": 9.101409101409102e-07,
      "loss": 0.0622,
      "step": 4417
    },
    {
      "epoch": 4.592515592515593,
      "grad_norm": 0.25829172134399414,
      "learning_rate": 9.078309078309079e-07,
      "loss": 0.0048,
      "step": 4418
    },
    {
      "epoch": 4.593555093555094,
      "grad_norm": 6.955380916595459,
      "learning_rate": 9.055209055209056e-07,
      "loss": 0.1169,
      "step": 4419
    },
    {
      "epoch": 4.594594594594595,
      "grad_norm": 2.080986976623535,
      "learning_rate": 9.032109032109033e-07,
      "loss": 0.0479,
      "step": 4420
    },
    {
      "epoch": 4.595634095634096,
      "grad_norm": 19.40532875061035,
      "learning_rate": 9.00900900900901e-07,
      "loss": 0.5309,
      "step": 4421
    },
    {
      "epoch": 4.596673596673597,
      "grad_norm": 0.6829842925071716,
      "learning_rate": 8.985908985908986e-07,
      "loss": 0.0135,
      "step": 4422
    },
    {
      "epoch": 4.597713097713098,
      "grad_norm": 0.22129260003566742,
      "learning_rate": 8.962808962808964e-07,
      "loss": 0.0056,
      "step": 4423
    },
    {
      "epoch": 4.598752598752599,
      "grad_norm": 5.869988918304443,
      "learning_rate": 8.939708939708941e-07,
      "loss": 0.0672,
      "step": 4424
    },
    {
      "epoch": 4.5997920997921,
      "grad_norm": 9.109978675842285,
      "learning_rate": 8.916608916608917e-07,
      "loss": 0.4428,
      "step": 4425
    },
    {
      "epoch": 4.600831600831601,
      "grad_norm": 3.746086359024048,
      "learning_rate": 8.893508893508894e-07,
      "loss": 0.2627,
      "step": 4426
    },
    {
      "epoch": 4.601871101871102,
      "grad_norm": 0.024450652301311493,
      "learning_rate": 8.870408870408871e-07,
      "loss": 0.0004,
      "step": 4427
    },
    {
      "epoch": 4.602910602910603,
      "grad_norm": 0.3670468330383301,
      "learning_rate": 8.847308847308848e-07,
      "loss": 0.0115,
      "step": 4428
    },
    {
      "epoch": 4.603950103950104,
      "grad_norm": 0.1896979957818985,
      "learning_rate": 8.824208824208826e-07,
      "loss": 0.0047,
      "step": 4429
    },
    {
      "epoch": 4.604989604989605,
      "grad_norm": 5.6438822746276855,
      "learning_rate": 8.801108801108801e-07,
      "loss": 0.5649,
      "step": 4430
    },
    {
      "epoch": 4.606029106029106,
      "grad_norm": 0.4361482858657837,
      "learning_rate": 8.778008778008779e-07,
      "loss": 0.0154,
      "step": 4431
    },
    {
      "epoch": 4.607068607068607,
      "grad_norm": 0.12333600968122482,
      "learning_rate": 8.754908754908756e-07,
      "loss": 0.0031,
      "step": 4432
    },
    {
      "epoch": 4.608108108108108,
      "grad_norm": 1.3953226804733276,
      "learning_rate": 8.731808731808733e-07,
      "loss": 0.0636,
      "step": 4433
    },
    {
      "epoch": 4.609147609147609,
      "grad_norm": 0.24091799557209015,
      "learning_rate": 8.708708708708709e-07,
      "loss": 0.0075,
      "step": 4434
    },
    {
      "epoch": 4.61018711018711,
      "grad_norm": 0.6747987270355225,
      "learning_rate": 8.685608685608687e-07,
      "loss": 0.014,
      "step": 4435
    },
    {
      "epoch": 4.611226611226611,
      "grad_norm": 2.1281895637512207,
      "learning_rate": 8.662508662508663e-07,
      "loss": 0.0412,
      "step": 4436
    },
    {
      "epoch": 4.612266112266112,
      "grad_norm": 1.3151452541351318,
      "learning_rate": 8.639408639408641e-07,
      "loss": 0.2408,
      "step": 4437
    },
    {
      "epoch": 4.613305613305613,
      "grad_norm": 5.143826961517334,
      "learning_rate": 8.616308616308617e-07,
      "loss": 0.1529,
      "step": 4438
    },
    {
      "epoch": 4.614345114345114,
      "grad_norm": 15.18270206451416,
      "learning_rate": 8.593208593208593e-07,
      "loss": 1.7764,
      "step": 4439
    },
    {
      "epoch": 4.615384615384615,
      "grad_norm": 6.709432125091553,
      "learning_rate": 8.570108570108571e-07,
      "loss": 0.1883,
      "step": 4440
    },
    {
      "epoch": 4.616424116424117,
      "grad_norm": 0.1298711895942688,
      "learning_rate": 8.547008547008548e-07,
      "loss": 0.0029,
      "step": 4441
    },
    {
      "epoch": 4.617463617463617,
      "grad_norm": 8.232173919677734,
      "learning_rate": 8.523908523908525e-07,
      "loss": 0.6129,
      "step": 4442
    },
    {
      "epoch": 4.618503118503119,
      "grad_norm": 1.6797841787338257,
      "learning_rate": 8.5008085008085e-07,
      "loss": 0.0228,
      "step": 4443
    },
    {
      "epoch": 4.61954261954262,
      "grad_norm": 0.4660455286502838,
      "learning_rate": 8.477708477708478e-07,
      "loss": 0.0106,
      "step": 4444
    },
    {
      "epoch": 4.620582120582121,
      "grad_norm": 0.5190257430076599,
      "learning_rate": 8.454608454608456e-07,
      "loss": 0.0069,
      "step": 4445
    },
    {
      "epoch": 4.621621621621622,
      "grad_norm": 4.557394027709961,
      "learning_rate": 8.431508431508432e-07,
      "loss": 0.1984,
      "step": 4446
    },
    {
      "epoch": 4.622661122661123,
      "grad_norm": 3.3669612407684326,
      "learning_rate": 8.40840840840841e-07,
      "loss": 0.0821,
      "step": 4447
    },
    {
      "epoch": 4.623700623700624,
      "grad_norm": 7.057306289672852,
      "learning_rate": 8.385308385308386e-07,
      "loss": 0.1986,
      "step": 4448
    },
    {
      "epoch": 4.624740124740125,
      "grad_norm": 6.849547386169434,
      "learning_rate": 8.362208362208362e-07,
      "loss": 0.6603,
      "step": 4449
    },
    {
      "epoch": 4.625779625779626,
      "grad_norm": 1.4122143983840942,
      "learning_rate": 8.33910833910834e-07,
      "loss": 0.0221,
      "step": 4450
    },
    {
      "epoch": 4.626819126819127,
      "grad_norm": 0.011345437727868557,
      "learning_rate": 8.316008316008318e-07,
      "loss": 0.0002,
      "step": 4451
    },
    {
      "epoch": 4.627858627858628,
      "grad_norm": 2.111825942993164,
      "learning_rate": 8.292908292908293e-07,
      "loss": 0.0669,
      "step": 4452
    },
    {
      "epoch": 4.628898128898129,
      "grad_norm": 0.4430828094482422,
      "learning_rate": 8.26980826980827e-07,
      "loss": 0.008,
      "step": 4453
    },
    {
      "epoch": 4.62993762993763,
      "grad_norm": 4.777477264404297,
      "learning_rate": 8.246708246708247e-07,
      "loss": 0.3356,
      "step": 4454
    },
    {
      "epoch": 4.630977130977131,
      "grad_norm": 10.0228910446167,
      "learning_rate": 8.223608223608224e-07,
      "loss": 0.2197,
      "step": 4455
    },
    {
      "epoch": 4.632016632016632,
      "grad_norm": 9.412625312805176,
      "learning_rate": 8.200508200508202e-07,
      "loss": 0.6145,
      "step": 4456
    },
    {
      "epoch": 4.633056133056133,
      "grad_norm": 2.350175619125366,
      "learning_rate": 8.177408177408177e-07,
      "loss": 0.0595,
      "step": 4457
    },
    {
      "epoch": 4.634095634095634,
      "grad_norm": 0.00910639762878418,
      "learning_rate": 8.154308154308155e-07,
      "loss": 0.0002,
      "step": 4458
    },
    {
      "epoch": 4.635135135135135,
      "grad_norm": 3.016007900238037,
      "learning_rate": 8.131208131208132e-07,
      "loss": 0.0865,
      "step": 4459
    },
    {
      "epoch": 4.636174636174636,
      "grad_norm": 5.475106716156006,
      "learning_rate": 8.108108108108109e-07,
      "loss": 0.1247,
      "step": 4460
    },
    {
      "epoch": 4.637214137214137,
      "grad_norm": 2.1070916652679443,
      "learning_rate": 8.085008085008085e-07,
      "loss": 0.0539,
      "step": 4461
    },
    {
      "epoch": 4.638253638253638,
      "grad_norm": 8.32010555267334,
      "learning_rate": 8.061908061908062e-07,
      "loss": 0.2201,
      "step": 4462
    },
    {
      "epoch": 4.639293139293139,
      "grad_norm": 11.642264366149902,
      "learning_rate": 8.038808038808039e-07,
      "loss": 0.3801,
      "step": 4463
    },
    {
      "epoch": 4.64033264033264,
      "grad_norm": 1.360060691833496,
      "learning_rate": 8.015708015708017e-07,
      "loss": 0.0294,
      "step": 4464
    },
    {
      "epoch": 4.641372141372141,
      "grad_norm": 3.858584403991699,
      "learning_rate": 7.992607992607993e-07,
      "loss": 0.0837,
      "step": 4465
    },
    {
      "epoch": 4.642411642411642,
      "grad_norm": 0.033442914485931396,
      "learning_rate": 7.96950796950797e-07,
      "loss": 0.0006,
      "step": 4466
    },
    {
      "epoch": 4.643451143451143,
      "grad_norm": 2.088732957839966,
      "learning_rate": 7.946407946407947e-07,
      "loss": 0.0696,
      "step": 4467
    },
    {
      "epoch": 4.644490644490644,
      "grad_norm": 14.464248657226562,
      "learning_rate": 7.923307923307924e-07,
      "loss": 0.8585,
      "step": 4468
    },
    {
      "epoch": 4.645530145530145,
      "grad_norm": 9.30983829498291,
      "learning_rate": 7.900207900207901e-07,
      "loss": 0.8977,
      "step": 4469
    },
    {
      "epoch": 4.646569646569646,
      "grad_norm": 0.00840518344193697,
      "learning_rate": 7.877107877107878e-07,
      "loss": 0.0003,
      "step": 4470
    },
    {
      "epoch": 4.647609147609147,
      "grad_norm": 0.9638694524765015,
      "learning_rate": 7.854007854007854e-07,
      "loss": 0.0188,
      "step": 4471
    },
    {
      "epoch": 4.648648648648649,
      "grad_norm": 9.1583833694458,
      "learning_rate": 7.830907830907832e-07,
      "loss": 0.3321,
      "step": 4472
    },
    {
      "epoch": 4.649688149688149,
      "grad_norm": 0.06451886147260666,
      "learning_rate": 7.807807807807808e-07,
      "loss": 0.0016,
      "step": 4473
    },
    {
      "epoch": 4.650727650727651,
      "grad_norm": 0.4256269335746765,
      "learning_rate": 7.784707784707786e-07,
      "loss": 0.0079,
      "step": 4474
    },
    {
      "epoch": 4.651767151767151,
      "grad_norm": 8.92230224609375,
      "learning_rate": 7.761607761607762e-07,
      "loss": 0.4972,
      "step": 4475
    },
    {
      "epoch": 4.652806652806653,
      "grad_norm": 0.922160267829895,
      "learning_rate": 7.738507738507739e-07,
      "loss": 0.0193,
      "step": 4476
    },
    {
      "epoch": 4.653846153846154,
      "grad_norm": 1.281999111175537,
      "learning_rate": 7.715407715407716e-07,
      "loss": 0.0445,
      "step": 4477
    },
    {
      "epoch": 4.654885654885655,
      "grad_norm": 1.8824506998062134,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.0331,
      "step": 4478
    },
    {
      "epoch": 4.655925155925156,
      "grad_norm": 5.897076606750488,
      "learning_rate": 7.669207669207669e-07,
      "loss": 0.1943,
      "step": 4479
    },
    {
      "epoch": 4.656964656964657,
      "grad_norm": 4.859743118286133,
      "learning_rate": 7.646107646107647e-07,
      "loss": 0.1367,
      "step": 4480
    },
    {
      "epoch": 4.658004158004158,
      "grad_norm": 0.13624556362628937,
      "learning_rate": 7.623007623007623e-07,
      "loss": 0.0035,
      "step": 4481
    },
    {
      "epoch": 4.659043659043659,
      "grad_norm": 2.1516993045806885,
      "learning_rate": 7.599907599907601e-07,
      "loss": 0.1007,
      "step": 4482
    },
    {
      "epoch": 4.66008316008316,
      "grad_norm": 0.3950836658477783,
      "learning_rate": 7.576807576807578e-07,
      "loss": 0.0109,
      "step": 4483
    },
    {
      "epoch": 4.661122661122661,
      "grad_norm": 4.261602401733398,
      "learning_rate": 7.553707553707554e-07,
      "loss": 0.195,
      "step": 4484
    },
    {
      "epoch": 4.662162162162162,
      "grad_norm": 11.667570114135742,
      "learning_rate": 7.530607530607531e-07,
      "loss": 0.7816,
      "step": 4485
    },
    {
      "epoch": 4.663201663201663,
      "grad_norm": 0.5048899054527283,
      "learning_rate": 7.507507507507509e-07,
      "loss": 0.0096,
      "step": 4486
    },
    {
      "epoch": 4.664241164241164,
      "grad_norm": 0.7986839413642883,
      "learning_rate": 7.484407484407485e-07,
      "loss": 0.0319,
      "step": 4487
    },
    {
      "epoch": 4.665280665280665,
      "grad_norm": 6.918520927429199,
      "learning_rate": 7.461307461307461e-07,
      "loss": 0.254,
      "step": 4488
    },
    {
      "epoch": 4.666320166320166,
      "grad_norm": 0.032249923795461655,
      "learning_rate": 7.438207438207438e-07,
      "loss": 0.0007,
      "step": 4489
    },
    {
      "epoch": 4.667359667359667,
      "grad_norm": 0.021945137530565262,
      "learning_rate": 7.415107415107416e-07,
      "loss": 0.0004,
      "step": 4490
    },
    {
      "epoch": 4.668399168399168,
      "grad_norm": 12.442795753479004,
      "learning_rate": 7.392007392007393e-07,
      "loss": 0.7439,
      "step": 4491
    },
    {
      "epoch": 4.669438669438669,
      "grad_norm": 2.157111883163452,
      "learning_rate": 7.36890736890737e-07,
      "loss": 0.0531,
      "step": 4492
    },
    {
      "epoch": 4.67047817047817,
      "grad_norm": 0.023635277524590492,
      "learning_rate": 7.345807345807346e-07,
      "loss": 0.0006,
      "step": 4493
    },
    {
      "epoch": 4.671517671517671,
      "grad_norm": 5.190340995788574,
      "learning_rate": 7.322707322707323e-07,
      "loss": 0.1485,
      "step": 4494
    },
    {
      "epoch": 4.672557172557172,
      "grad_norm": 0.08936876803636551,
      "learning_rate": 7.2996072996073e-07,
      "loss": 0.0018,
      "step": 4495
    },
    {
      "epoch": 4.673596673596673,
      "grad_norm": 7.4356465339660645,
      "learning_rate": 7.276507276507278e-07,
      "loss": 0.313,
      "step": 4496
    },
    {
      "epoch": 4.674636174636174,
      "grad_norm": 9.407391548156738,
      "learning_rate": 7.253407253407253e-07,
      "loss": 0.7526,
      "step": 4497
    },
    {
      "epoch": 4.675675675675675,
      "grad_norm": 1.671047568321228,
      "learning_rate": 7.23030723030723e-07,
      "loss": 0.0295,
      "step": 4498
    },
    {
      "epoch": 4.6767151767151764,
      "grad_norm": 9.059718132019043,
      "learning_rate": 7.207207207207208e-07,
      "loss": 0.9281,
      "step": 4499
    },
    {
      "epoch": 4.6777546777546775,
      "grad_norm": 0.3992685079574585,
      "learning_rate": 7.184107184107185e-07,
      "loss": 0.0069,
      "step": 4500
    },
    {
      "epoch": 4.6787941787941785,
      "grad_norm": 10.279568672180176,
      "learning_rate": 7.161007161007162e-07,
      "loss": 0.6143,
      "step": 4501
    },
    {
      "epoch": 4.6798336798336795,
      "grad_norm": 3.262343168258667,
      "learning_rate": 7.137907137907138e-07,
      "loss": 0.3237,
      "step": 4502
    },
    {
      "epoch": 4.6808731808731805,
      "grad_norm": 3.7217352390289307,
      "learning_rate": 7.114807114807115e-07,
      "loss": 0.1122,
      "step": 4503
    },
    {
      "epoch": 4.6819126819126815,
      "grad_norm": 0.7044948935508728,
      "learning_rate": 7.091707091707092e-07,
      "loss": 0.0217,
      "step": 4504
    },
    {
      "epoch": 4.682952182952183,
      "grad_norm": 5.225720405578613,
      "learning_rate": 7.06860706860707e-07,
      "loss": 0.2687,
      "step": 4505
    },
    {
      "epoch": 4.6839916839916835,
      "grad_norm": 3.316580295562744,
      "learning_rate": 7.045507045507045e-07,
      "loss": 0.1959,
      "step": 4506
    },
    {
      "epoch": 4.685031185031185,
      "grad_norm": 16.33987045288086,
      "learning_rate": 7.022407022407023e-07,
      "loss": 1.5211,
      "step": 4507
    },
    {
      "epoch": 4.686070686070686,
      "grad_norm": 7.938897132873535,
      "learning_rate": 6.999306999306999e-07,
      "loss": 0.2641,
      "step": 4508
    },
    {
      "epoch": 4.6871101871101875,
      "grad_norm": 0.2571137845516205,
      "learning_rate": 6.976206976206977e-07,
      "loss": 0.0094,
      "step": 4509
    },
    {
      "epoch": 4.6881496881496885,
      "grad_norm": 6.694173336029053,
      "learning_rate": 6.953106953106955e-07,
      "loss": 0.6494,
      "step": 4510
    },
    {
      "epoch": 4.6891891891891895,
      "grad_norm": 0.22093908488750458,
      "learning_rate": 6.93000693000693e-07,
      "loss": 0.0045,
      "step": 4511
    },
    {
      "epoch": 4.6902286902286905,
      "grad_norm": 0.020347382873296738,
      "learning_rate": 6.906906906906907e-07,
      "loss": 0.0003,
      "step": 4512
    },
    {
      "epoch": 4.6912681912681915,
      "grad_norm": 5.891967296600342,
      "learning_rate": 6.883806883806885e-07,
      "loss": 0.3368,
      "step": 4513
    },
    {
      "epoch": 4.6923076923076925,
      "grad_norm": 3.0930402278900146,
      "learning_rate": 6.860706860706861e-07,
      "loss": 0.0542,
      "step": 4514
    },
    {
      "epoch": 4.6933471933471935,
      "grad_norm": 11.057426452636719,
      "learning_rate": 6.837606837606839e-07,
      "loss": 0.6941,
      "step": 4515
    },
    {
      "epoch": 4.6943866943866945,
      "grad_norm": 0.2353089153766632,
      "learning_rate": 6.814506814506814e-07,
      "loss": 0.0044,
      "step": 4516
    },
    {
      "epoch": 4.6954261954261955,
      "grad_norm": 0.016913872212171555,
      "learning_rate": 6.791406791406792e-07,
      "loss": 0.0005,
      "step": 4517
    },
    {
      "epoch": 4.696465696465697,
      "grad_norm": 3.098769187927246,
      "learning_rate": 6.768306768306769e-07,
      "loss": 0.0645,
      "step": 4518
    },
    {
      "epoch": 4.697505197505198,
      "grad_norm": 0.6526622772216797,
      "learning_rate": 6.745206745206746e-07,
      "loss": 0.0176,
      "step": 4519
    },
    {
      "epoch": 4.698544698544699,
      "grad_norm": 1.5812019109725952,
      "learning_rate": 6.722106722106722e-07,
      "loss": 0.0398,
      "step": 4520
    },
    {
      "epoch": 4.6995841995842,
      "grad_norm": 0.003646696452051401,
      "learning_rate": 6.6990066990067e-07,
      "loss": 0.0001,
      "step": 4521
    },
    {
      "epoch": 4.700623700623701,
      "grad_norm": 1.0215805768966675,
      "learning_rate": 6.675906675906676e-07,
      "loss": 0.0511,
      "step": 4522
    },
    {
      "epoch": 4.701663201663202,
      "grad_norm": 2.810720682144165,
      "learning_rate": 6.652806652806654e-07,
      "loss": 0.0212,
      "step": 4523
    },
    {
      "epoch": 4.702702702702703,
      "grad_norm": 2.5942187309265137,
      "learning_rate": 6.62970662970663e-07,
      "loss": 0.043,
      "step": 4524
    },
    {
      "epoch": 4.703742203742204,
      "grad_norm": 0.024856634438037872,
      "learning_rate": 6.606606606606607e-07,
      "loss": 0.0006,
      "step": 4525
    },
    {
      "epoch": 4.704781704781705,
      "grad_norm": 8.373519897460938,
      "learning_rate": 6.583506583506584e-07,
      "loss": 0.7647,
      "step": 4526
    },
    {
      "epoch": 4.705821205821206,
      "grad_norm": 0.19453732669353485,
      "learning_rate": 6.560406560406561e-07,
      "loss": 0.0029,
      "step": 4527
    },
    {
      "epoch": 4.706860706860707,
      "grad_norm": 0.9466361999511719,
      "learning_rate": 6.537306537306538e-07,
      "loss": 0.0188,
      "step": 4528
    },
    {
      "epoch": 4.707900207900208,
      "grad_norm": 0.2725234627723694,
      "learning_rate": 6.514206514206515e-07,
      "loss": 0.0052,
      "step": 4529
    },
    {
      "epoch": 4.708939708939709,
      "grad_norm": 5.90593957901001,
      "learning_rate": 6.491106491106491e-07,
      "loss": 0.2407,
      "step": 4530
    },
    {
      "epoch": 4.70997920997921,
      "grad_norm": 8.032899856567383,
      "learning_rate": 6.468006468006469e-07,
      "loss": 0.4303,
      "step": 4531
    },
    {
      "epoch": 4.711018711018711,
      "grad_norm": 3.0624608993530273,
      "learning_rate": 6.444906444906446e-07,
      "loss": 0.0978,
      "step": 4532
    },
    {
      "epoch": 4.712058212058212,
      "grad_norm": 2.9606127738952637,
      "learning_rate": 6.421806421806423e-07,
      "loss": 0.0711,
      "step": 4533
    },
    {
      "epoch": 4.713097713097713,
      "grad_norm": 1.1739325523376465,
      "learning_rate": 6.398706398706399e-07,
      "loss": 0.2751,
      "step": 4534
    },
    {
      "epoch": 4.714137214137214,
      "grad_norm": 2.5979039669036865,
      "learning_rate": 6.375606375606376e-07,
      "loss": 0.0699,
      "step": 4535
    },
    {
      "epoch": 4.715176715176716,
      "grad_norm": 6.968671798706055,
      "learning_rate": 6.352506352506353e-07,
      "loss": 0.432,
      "step": 4536
    },
    {
      "epoch": 4.716216216216216,
      "grad_norm": 9.019438743591309,
      "learning_rate": 6.329406329406331e-07,
      "loss": 0.3449,
      "step": 4537
    },
    {
      "epoch": 4.717255717255718,
      "grad_norm": 5.718798637390137,
      "learning_rate": 6.306306306306306e-07,
      "loss": 0.2865,
      "step": 4538
    },
    {
      "epoch": 4.718295218295218,
      "grad_norm": 2.0874598026275635,
      "learning_rate": 6.283206283206284e-07,
      "loss": 0.0485,
      "step": 4539
    },
    {
      "epoch": 4.71933471933472,
      "grad_norm": 2.445598602294922,
      "learning_rate": 6.260106260106261e-07,
      "loss": 0.0968,
      "step": 4540
    },
    {
      "epoch": 4.720374220374221,
      "grad_norm": 9.9757661819458,
      "learning_rate": 6.237006237006238e-07,
      "loss": 0.5018,
      "step": 4541
    },
    {
      "epoch": 4.721413721413722,
      "grad_norm": 3.8065030574798584,
      "learning_rate": 6.213906213906215e-07,
      "loss": 0.0682,
      "step": 4542
    },
    {
      "epoch": 4.722453222453223,
      "grad_norm": 2.1041741371154785,
      "learning_rate": 6.190806190806191e-07,
      "loss": 0.0431,
      "step": 4543
    },
    {
      "epoch": 4.723492723492724,
      "grad_norm": 9.70471477508545,
      "learning_rate": 6.167706167706168e-07,
      "loss": 0.6736,
      "step": 4544
    },
    {
      "epoch": 4.724532224532225,
      "grad_norm": 4.165168285369873,
      "learning_rate": 6.144606144606146e-07,
      "loss": 0.201,
      "step": 4545
    },
    {
      "epoch": 4.725571725571726,
      "grad_norm": 1.6179542541503906,
      "learning_rate": 6.121506121506121e-07,
      "loss": 0.0314,
      "step": 4546
    },
    {
      "epoch": 4.726611226611227,
      "grad_norm": 6.723485946655273,
      "learning_rate": 6.098406098406099e-07,
      "loss": 0.2245,
      "step": 4547
    },
    {
      "epoch": 4.727650727650728,
      "grad_norm": 9.48215103149414,
      "learning_rate": 6.075306075306076e-07,
      "loss": 1.1988,
      "step": 4548
    },
    {
      "epoch": 4.728690228690229,
      "grad_norm": 2.4443297386169434,
      "learning_rate": 6.052206052206053e-07,
      "loss": 0.0589,
      "step": 4549
    },
    {
      "epoch": 4.72972972972973,
      "grad_norm": 0.08004553616046906,
      "learning_rate": 6.02910602910603e-07,
      "loss": 0.0017,
      "step": 4550
    },
    {
      "epoch": 4.730769230769231,
      "grad_norm": 6.5797200202941895,
      "learning_rate": 6.006006006006006e-07,
      "loss": 0.5225,
      "step": 4551
    },
    {
      "epoch": 4.731808731808732,
      "grad_norm": 1.0097525119781494,
      "learning_rate": 5.982905982905984e-07,
      "loss": 0.0388,
      "step": 4552
    },
    {
      "epoch": 4.732848232848233,
      "grad_norm": 0.07979341596364975,
      "learning_rate": 5.95980595980596e-07,
      "loss": 0.0026,
      "step": 4553
    },
    {
      "epoch": 4.733887733887734,
      "grad_norm": 9.43011474609375,
      "learning_rate": 5.936705936705937e-07,
      "loss": 0.4164,
      "step": 4554
    },
    {
      "epoch": 4.734927234927235,
      "grad_norm": 7.376195430755615,
      "learning_rate": 5.913605913605914e-07,
      "loss": 0.4523,
      "step": 4555
    },
    {
      "epoch": 4.735966735966736,
      "grad_norm": 1.2433668375015259,
      "learning_rate": 5.890505890505891e-07,
      "loss": 0.0614,
      "step": 4556
    },
    {
      "epoch": 4.737006237006237,
      "grad_norm": 4.30912971496582,
      "learning_rate": 5.867405867405867e-07,
      "loss": 0.1076,
      "step": 4557
    },
    {
      "epoch": 4.738045738045738,
      "grad_norm": 0.028791455551981926,
      "learning_rate": 5.844305844305845e-07,
      "loss": 0.0007,
      "step": 4558
    },
    {
      "epoch": 4.739085239085239,
      "grad_norm": 0.13918457925319672,
      "learning_rate": 5.821205821205821e-07,
      "loss": 0.0031,
      "step": 4559
    },
    {
      "epoch": 4.74012474012474,
      "grad_norm": 3.475895643234253,
      "learning_rate": 5.798105798105798e-07,
      "loss": 0.3015,
      "step": 4560
    },
    {
      "epoch": 4.741164241164241,
      "grad_norm": 0.010119406506419182,
      "learning_rate": 5.775005775005776e-07,
      "loss": 0.0003,
      "step": 4561
    },
    {
      "epoch": 4.742203742203742,
      "grad_norm": 1.498502492904663,
      "learning_rate": 5.751905751905752e-07,
      "loss": 0.0301,
      "step": 4562
    },
    {
      "epoch": 4.743243243243243,
      "grad_norm": 9.394128799438477,
      "learning_rate": 5.728805728805729e-07,
      "loss": 0.4929,
      "step": 4563
    },
    {
      "epoch": 4.744282744282744,
      "grad_norm": 0.28972873091697693,
      "learning_rate": 5.705705705705706e-07,
      "loss": 0.0087,
      "step": 4564
    },
    {
      "epoch": 4.745322245322245,
      "grad_norm": 1.398381233215332,
      "learning_rate": 5.682605682605683e-07,
      "loss": 0.0394,
      "step": 4565
    },
    {
      "epoch": 4.746361746361746,
      "grad_norm": 5.117181777954102,
      "learning_rate": 5.65950565950566e-07,
      "loss": 0.2151,
      "step": 4566
    },
    {
      "epoch": 4.747401247401247,
      "grad_norm": 6.448148250579834,
      "learning_rate": 5.636405636405637e-07,
      "loss": 0.3025,
      "step": 4567
    },
    {
      "epoch": 4.748440748440748,
      "grad_norm": 0.009139902889728546,
      "learning_rate": 5.613305613305614e-07,
      "loss": 0.0002,
      "step": 4568
    },
    {
      "epoch": 4.74948024948025,
      "grad_norm": 2.174379825592041,
      "learning_rate": 5.590205590205591e-07,
      "loss": 0.0265,
      "step": 4569
    },
    {
      "epoch": 4.75051975051975,
      "grad_norm": 6.330443382263184,
      "learning_rate": 5.567105567105567e-07,
      "loss": 0.1853,
      "step": 4570
    },
    {
      "epoch": 4.751559251559252,
      "grad_norm": 1.084800124168396,
      "learning_rate": 5.544005544005544e-07,
      "loss": 0.0169,
      "step": 4571
    },
    {
      "epoch": 4.752598752598753,
      "grad_norm": 10.983011245727539,
      "learning_rate": 5.520905520905522e-07,
      "loss": 0.3542,
      "step": 4572
    },
    {
      "epoch": 4.753638253638254,
      "grad_norm": 7.211549758911133,
      "learning_rate": 5.497805497805498e-07,
      "loss": 0.2281,
      "step": 4573
    },
    {
      "epoch": 4.754677754677755,
      "grad_norm": 0.026495108380913734,
      "learning_rate": 5.474705474705475e-07,
      "loss": 0.0007,
      "step": 4574
    },
    {
      "epoch": 4.755717255717256,
      "grad_norm": 1.1615225076675415,
      "learning_rate": 5.451605451605452e-07,
      "loss": 0.0235,
      "step": 4575
    },
    {
      "epoch": 4.756756756756757,
      "grad_norm": 0.4452331066131592,
      "learning_rate": 5.428505428505429e-07,
      "loss": 0.0107,
      "step": 4576
    },
    {
      "epoch": 4.757796257796258,
      "grad_norm": 0.0032271817326545715,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.0001,
      "step": 4577
    },
    {
      "epoch": 4.758835758835759,
      "grad_norm": 5.279394626617432,
      "learning_rate": 5.382305382305382e-07,
      "loss": 0.3176,
      "step": 4578
    },
    {
      "epoch": 4.75987525987526,
      "grad_norm": 10.044146537780762,
      "learning_rate": 5.35920535920536e-07,
      "loss": 0.4006,
      "step": 4579
    },
    {
      "epoch": 4.760914760914761,
      "grad_norm": 2.109079599380493,
      "learning_rate": 5.336105336105337e-07,
      "loss": 0.0414,
      "step": 4580
    },
    {
      "epoch": 4.761954261954262,
      "grad_norm": 3.126941204071045,
      "learning_rate": 5.313005313005313e-07,
      "loss": 0.0685,
      "step": 4581
    },
    {
      "epoch": 4.762993762993763,
      "grad_norm": 3.8069844245910645,
      "learning_rate": 5.28990528990529e-07,
      "loss": 0.1809,
      "step": 4582
    },
    {
      "epoch": 4.764033264033264,
      "grad_norm": 2.9434378147125244,
      "learning_rate": 5.266805266805268e-07,
      "loss": 0.252,
      "step": 4583
    },
    {
      "epoch": 4.765072765072765,
      "grad_norm": 0.6716585755348206,
      "learning_rate": 5.243705243705244e-07,
      "loss": 0.0158,
      "step": 4584
    },
    {
      "epoch": 4.766112266112266,
      "grad_norm": 5.513317108154297,
      "learning_rate": 5.220605220605221e-07,
      "loss": 0.1182,
      "step": 4585
    },
    {
      "epoch": 4.767151767151767,
      "grad_norm": 3.289609670639038,
      "learning_rate": 5.197505197505199e-07,
      "loss": 0.0477,
      "step": 4586
    },
    {
      "epoch": 4.768191268191268,
      "grad_norm": 0.6133396625518799,
      "learning_rate": 5.174405174405175e-07,
      "loss": 0.0119,
      "step": 4587
    },
    {
      "epoch": 4.769230769230769,
      "grad_norm": 4.571869373321533,
      "learning_rate": 5.151305151305152e-07,
      "loss": 0.5846,
      "step": 4588
    },
    {
      "epoch": 4.77027027027027,
      "grad_norm": 8.87611198425293,
      "learning_rate": 5.128205128205128e-07,
      "loss": 1.162,
      "step": 4589
    },
    {
      "epoch": 4.771309771309771,
      "grad_norm": 4.447170257568359,
      "learning_rate": 5.105105105105106e-07,
      "loss": 0.3333,
      "step": 4590
    },
    {
      "epoch": 4.772349272349272,
      "grad_norm": 10.814187049865723,
      "learning_rate": 5.082005082005083e-07,
      "loss": 0.506,
      "step": 4591
    },
    {
      "epoch": 4.773388773388773,
      "grad_norm": 8.977923393249512,
      "learning_rate": 5.058905058905059e-07,
      "loss": 0.5942,
      "step": 4592
    },
    {
      "epoch": 4.774428274428274,
      "grad_norm": 8.605423927307129,
      "learning_rate": 5.035805035805036e-07,
      "loss": 0.2784,
      "step": 4593
    },
    {
      "epoch": 4.775467775467775,
      "grad_norm": 0.4176355004310608,
      "learning_rate": 5.012705012705014e-07,
      "loss": 0.0105,
      "step": 4594
    },
    {
      "epoch": 4.776507276507276,
      "grad_norm": 11.354757308959961,
      "learning_rate": 4.98960498960499e-07,
      "loss": 0.7986,
      "step": 4595
    },
    {
      "epoch": 4.777546777546777,
      "grad_norm": 2.4823195934295654,
      "learning_rate": 4.966504966504967e-07,
      "loss": 0.0429,
      "step": 4596
    },
    {
      "epoch": 4.778586278586278,
      "grad_norm": 0.0783335343003273,
      "learning_rate": 4.943404943404944e-07,
      "loss": 0.0017,
      "step": 4597
    },
    {
      "epoch": 4.779625779625779,
      "grad_norm": 0.4593014121055603,
      "learning_rate": 4.92030492030492e-07,
      "loss": 0.0121,
      "step": 4598
    },
    {
      "epoch": 4.78066528066528,
      "grad_norm": 3.2849011421203613,
      "learning_rate": 4.897204897204898e-07,
      "loss": 0.0756,
      "step": 4599
    },
    {
      "epoch": 4.781704781704782,
      "grad_norm": 8.635833740234375,
      "learning_rate": 4.874104874104874e-07,
      "loss": 0.2125,
      "step": 4600
    },
    {
      "epoch": 4.782744282744282,
      "grad_norm": 1.040634274482727,
      "learning_rate": 4.851004851004852e-07,
      "loss": 0.0258,
      "step": 4601
    },
    {
      "epoch": 4.783783783783784,
      "grad_norm": 0.10529931634664536,
      "learning_rate": 4.827904827904829e-07,
      "loss": 0.0019,
      "step": 4602
    },
    {
      "epoch": 4.784823284823284,
      "grad_norm": 7.396679401397705,
      "learning_rate": 4.804804804804805e-07,
      "loss": 0.7434,
      "step": 4603
    },
    {
      "epoch": 4.785862785862786,
      "grad_norm": 11.395238876342773,
      "learning_rate": 4.781704781704783e-07,
      "loss": 1.2236,
      "step": 4604
    },
    {
      "epoch": 4.786902286902287,
      "grad_norm": 0.009259616024792194,
      "learning_rate": 4.758604758604759e-07,
      "loss": 0.0002,
      "step": 4605
    },
    {
      "epoch": 4.787941787941788,
      "grad_norm": 7.3028788566589355,
      "learning_rate": 4.735504735504736e-07,
      "loss": 0.2805,
      "step": 4606
    },
    {
      "epoch": 4.788981288981289,
      "grad_norm": 0.01691441796720028,
      "learning_rate": 4.7124047124047127e-07,
      "loss": 0.0003,
      "step": 4607
    },
    {
      "epoch": 4.79002079002079,
      "grad_norm": 0.09052679687738419,
      "learning_rate": 4.68930468930469e-07,
      "loss": 0.0014,
      "step": 4608
    },
    {
      "epoch": 4.791060291060291,
      "grad_norm": 3.935920476913452,
      "learning_rate": 4.6662046662046664e-07,
      "loss": 0.0984,
      "step": 4609
    },
    {
      "epoch": 4.792099792099792,
      "grad_norm": 12.794758796691895,
      "learning_rate": 4.6431046431046436e-07,
      "loss": 0.9687,
      "step": 4610
    },
    {
      "epoch": 4.793139293139293,
      "grad_norm": 1.4812192916870117,
      "learning_rate": 4.6200046200046207e-07,
      "loss": 0.0423,
      "step": 4611
    },
    {
      "epoch": 4.794178794178794,
      "grad_norm": 2.92372989654541,
      "learning_rate": 4.5969045969045973e-07,
      "loss": 0.084,
      "step": 4612
    },
    {
      "epoch": 4.795218295218295,
      "grad_norm": 3.692225217819214,
      "learning_rate": 4.5738045738045745e-07,
      "loss": 0.0971,
      "step": 4613
    },
    {
      "epoch": 4.796257796257796,
      "grad_norm": 0.20043006539344788,
      "learning_rate": 4.550704550704551e-07,
      "loss": 0.0063,
      "step": 4614
    },
    {
      "epoch": 4.797297297297297,
      "grad_norm": 6.279969215393066,
      "learning_rate": 4.527604527604528e-07,
      "loss": 0.2496,
      "step": 4615
    },
    {
      "epoch": 4.798336798336798,
      "grad_norm": 0.7676145434379578,
      "learning_rate": 4.504504504504505e-07,
      "loss": 0.0142,
      "step": 4616
    },
    {
      "epoch": 4.799376299376299,
      "grad_norm": 0.9839637875556946,
      "learning_rate": 4.481404481404482e-07,
      "loss": 0.0507,
      "step": 4617
    },
    {
      "epoch": 4.8004158004158,
      "grad_norm": 1.1574064493179321,
      "learning_rate": 4.4583044583044586e-07,
      "loss": 0.2651,
      "step": 4618
    },
    {
      "epoch": 4.801455301455301,
      "grad_norm": 2.5498995780944824,
      "learning_rate": 4.435204435204436e-07,
      "loss": 0.074,
      "step": 4619
    },
    {
      "epoch": 4.802494802494802,
      "grad_norm": 11.176398277282715,
      "learning_rate": 4.412104412104413e-07,
      "loss": 1.3905,
      "step": 4620
    },
    {
      "epoch": 4.803534303534303,
      "grad_norm": NaN,
      "learning_rate": 4.3890043890043895e-07,
      "loss": 0.0692,
      "step": 4621
    },
    {
      "epoch": 4.8045738045738045,
      "grad_norm": 0.02660849131643772,
      "learning_rate": 4.3659043659043666e-07,
      "loss": 0.0005,
      "step": 4622
    },
    {
      "epoch": 4.8056133056133055,
      "grad_norm": 15.595175743103027,
      "learning_rate": 4.342804342804343e-07,
      "loss": 0.5628,
      "step": 4623
    },
    {
      "epoch": 4.8066528066528065,
      "grad_norm": 0.4597373902797699,
      "learning_rate": 4.3197043197043204e-07,
      "loss": 0.0093,
      "step": 4624
    },
    {
      "epoch": 4.8076923076923075,
      "grad_norm": 6.716993808746338,
      "learning_rate": 4.2966042966042965e-07,
      "loss": 0.3927,
      "step": 4625
    },
    {
      "epoch": 4.8087318087318085,
      "grad_norm": 3.250091314315796,
      "learning_rate": 4.273504273504274e-07,
      "loss": 0.1009,
      "step": 4626
    },
    {
      "epoch": 4.8097713097713095,
      "grad_norm": 9.529396057128906,
      "learning_rate": 4.25040425040425e-07,
      "loss": 1.1618,
      "step": 4627
    },
    {
      "epoch": 4.8108108108108105,
      "grad_norm": 0.3871168792247772,
      "learning_rate": 4.227304227304228e-07,
      "loss": 0.0091,
      "step": 4628
    },
    {
      "epoch": 4.8118503118503115,
      "grad_norm": 0.20890498161315918,
      "learning_rate": 4.204204204204205e-07,
      "loss": 0.0053,
      "step": 4629
    },
    {
      "epoch": 4.8128898128898125,
      "grad_norm": 0.05548560619354248,
      "learning_rate": 4.181104181104181e-07,
      "loss": 0.001,
      "step": 4630
    },
    {
      "epoch": 4.813929313929314,
      "grad_norm": 4.1424078941345215,
      "learning_rate": 4.158004158004159e-07,
      "loss": 0.4125,
      "step": 4631
    },
    {
      "epoch": 4.814968814968815,
      "grad_norm": 6.571180820465088,
      "learning_rate": 4.134904134904135e-07,
      "loss": 0.1227,
      "step": 4632
    },
    {
      "epoch": 4.8160083160083165,
      "grad_norm": 1.9572172164916992,
      "learning_rate": 4.111804111804112e-07,
      "loss": 0.1683,
      "step": 4633
    },
    {
      "epoch": 4.817047817047817,
      "grad_norm": 0.7159575819969177,
      "learning_rate": 4.0887040887040887e-07,
      "loss": 0.0164,
      "step": 4634
    },
    {
      "epoch": 4.8180873180873185,
      "grad_norm": 4.800737380981445,
      "learning_rate": 4.065604065604066e-07,
      "loss": 0.1449,
      "step": 4635
    },
    {
      "epoch": 4.8191268191268195,
      "grad_norm": 0.29823213815689087,
      "learning_rate": 4.0425040425040424e-07,
      "loss": 0.0079,
      "step": 4636
    },
    {
      "epoch": 4.8201663201663205,
      "grad_norm": 9.392901420593262,
      "learning_rate": 4.0194040194040195e-07,
      "loss": 0.4268,
      "step": 4637
    },
    {
      "epoch": 4.8212058212058215,
      "grad_norm": 0.028579136356711388,
      "learning_rate": 3.9963039963039967e-07,
      "loss": 0.0008,
      "step": 4638
    },
    {
      "epoch": 4.8222453222453225,
      "grad_norm": 2.9507510662078857,
      "learning_rate": 3.9732039732039733e-07,
      "loss": 0.0765,
      "step": 4639
    },
    {
      "epoch": 4.8232848232848236,
      "grad_norm": 7.703818321228027,
      "learning_rate": 3.9501039501039504e-07,
      "loss": 0.338,
      "step": 4640
    },
    {
      "epoch": 4.824324324324325,
      "grad_norm": 4.047659397125244,
      "learning_rate": 3.927003927003927e-07,
      "loss": 0.2549,
      "step": 4641
    },
    {
      "epoch": 4.825363825363826,
      "grad_norm": 0.6981263756752014,
      "learning_rate": 3.903903903903904e-07,
      "loss": 0.0229,
      "step": 4642
    },
    {
      "epoch": 4.826403326403327,
      "grad_norm": 4.9491777420043945,
      "learning_rate": 3.880803880803881e-07,
      "loss": 0.1437,
      "step": 4643
    },
    {
      "epoch": 4.827442827442828,
      "grad_norm": 1.2429845333099365,
      "learning_rate": 3.857703857703858e-07,
      "loss": 0.039,
      "step": 4644
    },
    {
      "epoch": 4.828482328482329,
      "grad_norm": 0.000703688245266676,
      "learning_rate": 3.8346038346038346e-07,
      "loss": 0.0,
      "step": 4645
    },
    {
      "epoch": 4.82952182952183,
      "grad_norm": 6.003587245941162,
      "learning_rate": 3.8115038115038117e-07,
      "loss": 0.2291,
      "step": 4646
    },
    {
      "epoch": 4.830561330561331,
      "grad_norm": 9.933477401733398,
      "learning_rate": 3.788403788403789e-07,
      "loss": 0.7847,
      "step": 4647
    },
    {
      "epoch": 4.831600831600832,
      "grad_norm": 14.56813907623291,
      "learning_rate": 3.7653037653037655e-07,
      "loss": 0.6007,
      "step": 4648
    },
    {
      "epoch": 4.832640332640333,
      "grad_norm": 4.413318157196045,
      "learning_rate": 3.7422037422037426e-07,
      "loss": 0.1072,
      "step": 4649
    },
    {
      "epoch": 4.833679833679834,
      "grad_norm": 1.5492770671844482,
      "learning_rate": 3.719103719103719e-07,
      "loss": 0.0357,
      "step": 4650
    },
    {
      "epoch": 4.834719334719335,
      "grad_norm": 9.484484672546387,
      "learning_rate": 3.6960036960036964e-07,
      "loss": 0.5245,
      "step": 4651
    },
    {
      "epoch": 4.835758835758836,
      "grad_norm": 0.01474912278354168,
      "learning_rate": 3.672903672903673e-07,
      "loss": 0.0004,
      "step": 4652
    },
    {
      "epoch": 4.836798336798337,
      "grad_norm": 10.440610885620117,
      "learning_rate": 3.64980364980365e-07,
      "loss": 1.6802,
      "step": 4653
    },
    {
      "epoch": 4.837837837837838,
      "grad_norm": 3.246966600418091,
      "learning_rate": 3.626703626703627e-07,
      "loss": 0.2942,
      "step": 4654
    },
    {
      "epoch": 4.838877338877339,
      "grad_norm": 8.040372848510742,
      "learning_rate": 3.603603603603604e-07,
      "loss": 0.2389,
      "step": 4655
    },
    {
      "epoch": 4.83991683991684,
      "grad_norm": 0.1538681983947754,
      "learning_rate": 3.580503580503581e-07,
      "loss": 0.0029,
      "step": 4656
    },
    {
      "epoch": 4.840956340956341,
      "grad_norm": 8.433476448059082,
      "learning_rate": 3.5574035574035576e-07,
      "loss": 0.7995,
      "step": 4657
    },
    {
      "epoch": 4.841995841995842,
      "grad_norm": 0.0016116875922307372,
      "learning_rate": 3.534303534303535e-07,
      "loss": 0.0,
      "step": 4658
    },
    {
      "epoch": 4.843035343035343,
      "grad_norm": 4.5516133308410645,
      "learning_rate": 3.5112035112035114e-07,
      "loss": 0.1731,
      "step": 4659
    },
    {
      "epoch": 4.844074844074844,
      "grad_norm": 0.5104297399520874,
      "learning_rate": 3.4881034881034885e-07,
      "loss": 0.0108,
      "step": 4660
    },
    {
      "epoch": 4.845114345114345,
      "grad_norm": 0.16368019580841064,
      "learning_rate": 3.465003465003465e-07,
      "loss": 0.0047,
      "step": 4661
    },
    {
      "epoch": 4.846153846153846,
      "grad_norm": 9.44383716583252,
      "learning_rate": 3.4419034419034423e-07,
      "loss": 0.2378,
      "step": 4662
    },
    {
      "epoch": 4.847193347193347,
      "grad_norm": 0.10227713733911514,
      "learning_rate": 3.4188034188034194e-07,
      "loss": 0.0021,
      "step": 4663
    },
    {
      "epoch": 4.848232848232849,
      "grad_norm": 5.760284423828125,
      "learning_rate": 3.395703395703396e-07,
      "loss": 0.2681,
      "step": 4664
    },
    {
      "epoch": 4.849272349272349,
      "grad_norm": 5.341785907745361,
      "learning_rate": 3.372603372603373e-07,
      "loss": 0.2131,
      "step": 4665
    },
    {
      "epoch": 4.850311850311851,
      "grad_norm": 7.600803852081299,
      "learning_rate": 3.34950334950335e-07,
      "loss": 0.3075,
      "step": 4666
    },
    {
      "epoch": 4.851351351351351,
      "grad_norm": 0.27108874917030334,
      "learning_rate": 3.326403326403327e-07,
      "loss": 0.0066,
      "step": 4667
    },
    {
      "epoch": 4.852390852390853,
      "grad_norm": 26.737104415893555,
      "learning_rate": 3.3033033033033036e-07,
      "loss": 0.524,
      "step": 4668
    },
    {
      "epoch": 4.853430353430354,
      "grad_norm": 1.14380943775177,
      "learning_rate": 3.2802032802032807e-07,
      "loss": 0.0217,
      "step": 4669
    },
    {
      "epoch": 4.854469854469855,
      "grad_norm": 5.919650077819824,
      "learning_rate": 3.2571032571032573e-07,
      "loss": 0.3197,
      "step": 4670
    },
    {
      "epoch": 4.855509355509356,
      "grad_norm": 3.418506622314453,
      "learning_rate": 3.2340032340032345e-07,
      "loss": 0.1859,
      "step": 4671
    },
    {
      "epoch": 4.856548856548857,
      "grad_norm": 6.070278167724609,
      "learning_rate": 3.2109032109032116e-07,
      "loss": 0.1995,
      "step": 4672
    },
    {
      "epoch": 4.857588357588358,
      "grad_norm": 0.038140926510095596,
      "learning_rate": 3.187803187803188e-07,
      "loss": 0.0008,
      "step": 4673
    },
    {
      "epoch": 4.858627858627859,
      "grad_norm": 0.01792663149535656,
      "learning_rate": 3.1647031647031654e-07,
      "loss": 0.0003,
      "step": 4674
    },
    {
      "epoch": 4.85966735966736,
      "grad_norm": 11.694622039794922,
      "learning_rate": 3.141603141603142e-07,
      "loss": 0.6092,
      "step": 4675
    },
    {
      "epoch": 4.860706860706861,
      "grad_norm": 0.8555666208267212,
      "learning_rate": 3.118503118503119e-07,
      "loss": 0.018,
      "step": 4676
    },
    {
      "epoch": 4.861746361746362,
      "grad_norm": 1.2930259704589844,
      "learning_rate": 3.0954030954030957e-07,
      "loss": 0.0328,
      "step": 4677
    },
    {
      "epoch": 4.862785862785863,
      "grad_norm": 0.6727796792984009,
      "learning_rate": 3.072303072303073e-07,
      "loss": 0.0152,
      "step": 4678
    },
    {
      "epoch": 4.863825363825364,
      "grad_norm": 0.08168227225542068,
      "learning_rate": 3.0492030492030495e-07,
      "loss": 0.0014,
      "step": 4679
    },
    {
      "epoch": 4.864864864864865,
      "grad_norm": 13.4307279586792,
      "learning_rate": 3.0261030261030266e-07,
      "loss": 1.2033,
      "step": 4680
    },
    {
      "epoch": 4.865904365904366,
      "grad_norm": 1.3976296186447144,
      "learning_rate": 3.003003003003003e-07,
      "loss": 0.0443,
      "step": 4681
    },
    {
      "epoch": 4.866943866943867,
      "grad_norm": 5.100250244140625,
      "learning_rate": 2.97990297990298e-07,
      "loss": 0.2364,
      "step": 4682
    },
    {
      "epoch": 4.867983367983368,
      "grad_norm": 1.120133399963379,
      "learning_rate": 2.956802956802957e-07,
      "loss": 0.0242,
      "step": 4683
    },
    {
      "epoch": 4.869022869022869,
      "grad_norm": 9.938117980957031,
      "learning_rate": 2.9337029337029336e-07,
      "loss": 0.483,
      "step": 4684
    },
    {
      "epoch": 4.87006237006237,
      "grad_norm": 0.07843750715255737,
      "learning_rate": 2.910602910602911e-07,
      "loss": 0.0017,
      "step": 4685
    },
    {
      "epoch": 4.871101871101871,
      "grad_norm": 4.453456401824951,
      "learning_rate": 2.887502887502888e-07,
      "loss": 0.113,
      "step": 4686
    },
    {
      "epoch": 4.872141372141372,
      "grad_norm": 1.9144773483276367,
      "learning_rate": 2.8644028644028645e-07,
      "loss": 0.0655,
      "step": 4687
    },
    {
      "epoch": 4.873180873180873,
      "grad_norm": 8.914125442504883,
      "learning_rate": 2.8413028413028416e-07,
      "loss": 0.7485,
      "step": 4688
    },
    {
      "epoch": 4.874220374220374,
      "grad_norm": 9.680907249450684,
      "learning_rate": 2.818202818202818e-07,
      "loss": 0.9015,
      "step": 4689
    },
    {
      "epoch": 4.875259875259875,
      "grad_norm": 5.181673526763916,
      "learning_rate": 2.7951027951027954e-07,
      "loss": 0.0536,
      "step": 4690
    },
    {
      "epoch": 4.876299376299376,
      "grad_norm": 7.63839864730835,
      "learning_rate": 2.772002772002772e-07,
      "loss": 0.5203,
      "step": 4691
    },
    {
      "epoch": 4.877338877338877,
      "grad_norm": 2.4442501068115234,
      "learning_rate": 2.748902748902749e-07,
      "loss": 0.069,
      "step": 4692
    },
    {
      "epoch": 4.878378378378378,
      "grad_norm": 11.12451457977295,
      "learning_rate": 2.725802725802726e-07,
      "loss": 0.8014,
      "step": 4693
    },
    {
      "epoch": 4.879417879417879,
      "grad_norm": 1.978690505027771,
      "learning_rate": 2.702702702702703e-07,
      "loss": 0.054,
      "step": 4694
    },
    {
      "epoch": 4.88045738045738,
      "grad_norm": 0.7453691959381104,
      "learning_rate": 2.67960267960268e-07,
      "loss": 0.0075,
      "step": 4695
    },
    {
      "epoch": 4.881496881496881,
      "grad_norm": 0.621982753276825,
      "learning_rate": 2.6565026565026567e-07,
      "loss": 0.014,
      "step": 4696
    },
    {
      "epoch": 4.882536382536383,
      "grad_norm": 1.5242571830749512,
      "learning_rate": 2.633402633402634e-07,
      "loss": 0.0337,
      "step": 4697
    },
    {
      "epoch": 4.883575883575883,
      "grad_norm": 7.722758769989014,
      "learning_rate": 2.6103026103026104e-07,
      "loss": 0.3065,
      "step": 4698
    },
    {
      "epoch": 4.884615384615385,
      "grad_norm": 9.232809066772461,
      "learning_rate": 2.5872025872025876e-07,
      "loss": 0.5869,
      "step": 4699
    },
    {
      "epoch": 4.885654885654886,
      "grad_norm": 7.493601322174072,
      "learning_rate": 2.564102564102564e-07,
      "loss": 0.7301,
      "step": 4700
    },
    {
      "epoch": 4.886694386694387,
      "grad_norm": 3.2477588653564453,
      "learning_rate": 2.5410025410025413e-07,
      "loss": 0.059,
      "step": 4701
    },
    {
      "epoch": 4.887733887733888,
      "grad_norm": 2.048398017883301,
      "learning_rate": 2.517902517902518e-07,
      "loss": 0.0386,
      "step": 4702
    },
    {
      "epoch": 4.888773388773389,
      "grad_norm": 10.010037422180176,
      "learning_rate": 2.494802494802495e-07,
      "loss": 0.5694,
      "step": 4703
    },
    {
      "epoch": 4.88981288981289,
      "grad_norm": 5.604851245880127,
      "learning_rate": 2.471702471702472e-07,
      "loss": 0.4441,
      "step": 4704
    },
    {
      "epoch": 4.890852390852391,
      "grad_norm": 1.5705468654632568,
      "learning_rate": 2.448602448602449e-07,
      "loss": 0.2935,
      "step": 4705
    },
    {
      "epoch": 4.891891891891892,
      "grad_norm": 8.666454315185547,
      "learning_rate": 2.425502425502426e-07,
      "loss": 0.9524,
      "step": 4706
    },
    {
      "epoch": 4.892931392931393,
      "grad_norm": 3.345090389251709,
      "learning_rate": 2.4024024024024026e-07,
      "loss": 0.096,
      "step": 4707
    },
    {
      "epoch": 4.893970893970894,
      "grad_norm": 0.24659332633018494,
      "learning_rate": 2.3793023793023795e-07,
      "loss": 0.0042,
      "step": 4708
    },
    {
      "epoch": 4.895010395010395,
      "grad_norm": 3.87328839302063,
      "learning_rate": 2.3562023562023563e-07,
      "loss": 0.1188,
      "step": 4709
    },
    {
      "epoch": 4.896049896049896,
      "grad_norm": 0.04412136226892471,
      "learning_rate": 2.3331023331023332e-07,
      "loss": 0.0009,
      "step": 4710
    },
    {
      "epoch": 4.897089397089397,
      "grad_norm": 0.9146865010261536,
      "learning_rate": 2.3100023100023104e-07,
      "loss": 0.0171,
      "step": 4711
    },
    {
      "epoch": 4.898128898128898,
      "grad_norm": 2.897141456604004,
      "learning_rate": 2.2869022869022872e-07,
      "loss": 0.0577,
      "step": 4712
    },
    {
      "epoch": 4.899168399168399,
      "grad_norm": 7.913065433502197,
      "learning_rate": 2.263802263802264e-07,
      "loss": 0.6139,
      "step": 4713
    },
    {
      "epoch": 4.9002079002079,
      "grad_norm": 21.14871597290039,
      "learning_rate": 2.240702240702241e-07,
      "loss": 1.5482,
      "step": 4714
    },
    {
      "epoch": 4.901247401247401,
      "grad_norm": 0.7635882496833801,
      "learning_rate": 2.217602217602218e-07,
      "loss": 0.0128,
      "step": 4715
    },
    {
      "epoch": 4.902286902286902,
      "grad_norm": 5.794342994689941,
      "learning_rate": 2.1945021945021948e-07,
      "loss": 0.1829,
      "step": 4716
    },
    {
      "epoch": 4.903326403326403,
      "grad_norm": 8.451240539550781,
      "learning_rate": 2.1714021714021716e-07,
      "loss": 0.1605,
      "step": 4717
    },
    {
      "epoch": 4.904365904365904,
      "grad_norm": 4.783844470977783,
      "learning_rate": 2.1483021483021482e-07,
      "loss": 0.0833,
      "step": 4718
    },
    {
      "epoch": 4.905405405405405,
      "grad_norm": 9.323697090148926,
      "learning_rate": 2.125202125202125e-07,
      "loss": 0.9744,
      "step": 4719
    },
    {
      "epoch": 4.906444906444906,
      "grad_norm": 0.004284221678972244,
      "learning_rate": 2.1021021021021025e-07,
      "loss": 0.0001,
      "step": 4720
    },
    {
      "epoch": 4.907484407484407,
      "grad_norm": 0.03907490149140358,
      "learning_rate": 2.0790020790020794e-07,
      "loss": 0.0011,
      "step": 4721
    },
    {
      "epoch": 4.908523908523908,
      "grad_norm": 1.798534631729126,
      "learning_rate": 2.055902055902056e-07,
      "loss": 0.0782,
      "step": 4722
    },
    {
      "epoch": 4.909563409563409,
      "grad_norm": 0.10509315133094788,
      "learning_rate": 2.032802032802033e-07,
      "loss": 0.0033,
      "step": 4723
    },
    {
      "epoch": 4.91060291060291,
      "grad_norm": 0.04313625395298004,
      "learning_rate": 2.0097020097020098e-07,
      "loss": 0.0008,
      "step": 4724
    },
    {
      "epoch": 4.911642411642411,
      "grad_norm": 3.6045427322387695,
      "learning_rate": 1.9866019866019867e-07,
      "loss": 0.0667,
      "step": 4725
    },
    {
      "epoch": 4.912681912681912,
      "grad_norm": 0.20354896783828735,
      "learning_rate": 1.9635019635019635e-07,
      "loss": 0.0038,
      "step": 4726
    },
    {
      "epoch": 4.913721413721413,
      "grad_norm": 3.984830856323242,
      "learning_rate": 1.9404019404019404e-07,
      "loss": 0.2169,
      "step": 4727
    },
    {
      "epoch": 4.914760914760915,
      "grad_norm": 4.478424072265625,
      "learning_rate": 1.9173019173019173e-07,
      "loss": 0.2199,
      "step": 4728
    },
    {
      "epoch": 4.915800415800415,
      "grad_norm": 0.11140380799770355,
      "learning_rate": 1.8942018942018944e-07,
      "loss": 0.0027,
      "step": 4729
    },
    {
      "epoch": 4.916839916839917,
      "grad_norm": 2.437990665435791,
      "learning_rate": 1.8711018711018713e-07,
      "loss": 0.2274,
      "step": 4730
    },
    {
      "epoch": 4.917879417879417,
      "grad_norm": 11.711197853088379,
      "learning_rate": 1.8480018480018482e-07,
      "loss": 1.0555,
      "step": 4731
    },
    {
      "epoch": 4.918918918918919,
      "grad_norm": 13.96945571899414,
      "learning_rate": 1.824901824901825e-07,
      "loss": 1.3861,
      "step": 4732
    },
    {
      "epoch": 4.91995841995842,
      "grad_norm": 8.046520233154297,
      "learning_rate": 1.801801801801802e-07,
      "loss": 0.5197,
      "step": 4733
    },
    {
      "epoch": 4.920997920997921,
      "grad_norm": 3.5241611003875732,
      "learning_rate": 1.7787017787017788e-07,
      "loss": 0.1781,
      "step": 4734
    },
    {
      "epoch": 4.922037422037422,
      "grad_norm": 0.1690087765455246,
      "learning_rate": 1.7556017556017557e-07,
      "loss": 0.0038,
      "step": 4735
    },
    {
      "epoch": 4.923076923076923,
      "grad_norm": 1.308184027671814,
      "learning_rate": 1.7325017325017326e-07,
      "loss": 0.0249,
      "step": 4736
    },
    {
      "epoch": 4.924116424116424,
      "grad_norm": 0.6014237999916077,
      "learning_rate": 1.7094017094017097e-07,
      "loss": 0.0142,
      "step": 4737
    },
    {
      "epoch": 4.925155925155925,
      "grad_norm": 3.1152658462524414,
      "learning_rate": 1.6863016863016866e-07,
      "loss": 0.0834,
      "step": 4738
    },
    {
      "epoch": 4.926195426195426,
      "grad_norm": 3.200228452682495,
      "learning_rate": 1.6632016632016635e-07,
      "loss": 0.0566,
      "step": 4739
    },
    {
      "epoch": 4.927234927234927,
      "grad_norm": 6.555817127227783,
      "learning_rate": 1.6401016401016403e-07,
      "loss": 0.1928,
      "step": 4740
    },
    {
      "epoch": 4.928274428274428,
      "grad_norm": 10.922245979309082,
      "learning_rate": 1.6170016170016172e-07,
      "loss": 1.2626,
      "step": 4741
    },
    {
      "epoch": 4.929313929313929,
      "grad_norm": 8.992452621459961,
      "learning_rate": 1.593901593901594e-07,
      "loss": 0.381,
      "step": 4742
    },
    {
      "epoch": 4.93035343035343,
      "grad_norm": 1.0346816778182983,
      "learning_rate": 1.570801570801571e-07,
      "loss": 0.0261,
      "step": 4743
    },
    {
      "epoch": 4.9313929313929314,
      "grad_norm": 0.0066374437883496284,
      "learning_rate": 1.5477015477015479e-07,
      "loss": 0.0002,
      "step": 4744
    },
    {
      "epoch": 4.9324324324324325,
      "grad_norm": 0.47477760910987854,
      "learning_rate": 1.5246015246015247e-07,
      "loss": 0.0115,
      "step": 4745
    },
    {
      "epoch": 4.9334719334719335,
      "grad_norm": 10.687585830688477,
      "learning_rate": 1.5015015015015016e-07,
      "loss": 1.1373,
      "step": 4746
    },
    {
      "epoch": 4.9345114345114345,
      "grad_norm": 7.225226879119873,
      "learning_rate": 1.4784014784014785e-07,
      "loss": 0.2633,
      "step": 4747
    },
    {
      "epoch": 4.9355509355509355,
      "grad_norm": 2.706840991973877,
      "learning_rate": 1.4553014553014554e-07,
      "loss": 0.0638,
      "step": 4748
    },
    {
      "epoch": 4.9365904365904365,
      "grad_norm": 4.533247470855713,
      "learning_rate": 1.4322014322014322e-07,
      "loss": 0.2218,
      "step": 4749
    },
    {
      "epoch": 4.9376299376299375,
      "grad_norm": 9.23600959777832,
      "learning_rate": 1.409101409101409e-07,
      "loss": 0.2363,
      "step": 4750
    },
    {
      "epoch": 4.9386694386694385,
      "grad_norm": 4.492886543273926,
      "learning_rate": 1.386001386001386e-07,
      "loss": 0.1644,
      "step": 4751
    },
    {
      "epoch": 4.9397089397089395,
      "grad_norm": 1.5420405864715576,
      "learning_rate": 1.362901362901363e-07,
      "loss": 0.2551,
      "step": 4752
    },
    {
      "epoch": 4.9407484407484406,
      "grad_norm": 2.434460163116455,
      "learning_rate": 1.33980133980134e-07,
      "loss": 0.0508,
      "step": 4753
    },
    {
      "epoch": 4.941787941787942,
      "grad_norm": 6.422155857086182,
      "learning_rate": 1.316701316701317e-07,
      "loss": 0.2384,
      "step": 4754
    },
    {
      "epoch": 4.942827442827443,
      "grad_norm": 0.10697789490222931,
      "learning_rate": 1.2936012936012938e-07,
      "loss": 0.0013,
      "step": 4755
    },
    {
      "epoch": 4.943866943866944,
      "grad_norm": 4.407161712646484,
      "learning_rate": 1.2705012705012707e-07,
      "loss": 0.2158,
      "step": 4756
    },
    {
      "epoch": 4.944906444906445,
      "grad_norm": 1.721889853477478,
      "learning_rate": 1.2474012474012475e-07,
      "loss": 0.0384,
      "step": 4757
    },
    {
      "epoch": 4.945945945945946,
      "grad_norm": 4.189394950866699,
      "learning_rate": 1.2243012243012244e-07,
      "loss": 0.1161,
      "step": 4758
    },
    {
      "epoch": 4.946985446985447,
      "grad_norm": 7.549504280090332,
      "learning_rate": 1.2012012012012013e-07,
      "loss": 0.2183,
      "step": 4759
    },
    {
      "epoch": 4.948024948024948,
      "grad_norm": 5.708642959594727,
      "learning_rate": 1.1781011781011782e-07,
      "loss": 0.4252,
      "step": 4760
    },
    {
      "epoch": 4.9490644490644495,
      "grad_norm": 2.221500873565674,
      "learning_rate": 1.1550011550011552e-07,
      "loss": 0.0502,
      "step": 4761
    },
    {
      "epoch": 4.95010395010395,
      "grad_norm": 0.014133865013718605,
      "learning_rate": 1.131901131901132e-07,
      "loss": 0.0003,
      "step": 4762
    },
    {
      "epoch": 4.951143451143452,
      "grad_norm": 0.6203381419181824,
      "learning_rate": 1.108801108801109e-07,
      "loss": 0.0177,
      "step": 4763
    },
    {
      "epoch": 4.952182952182953,
      "grad_norm": 8.374246597290039,
      "learning_rate": 1.0857010857010858e-07,
      "loss": 0.7584,
      "step": 4764
    },
    {
      "epoch": 4.953222453222454,
      "grad_norm": 5.350039482116699,
      "learning_rate": 1.0626010626010626e-07,
      "loss": 0.1403,
      "step": 4765
    },
    {
      "epoch": 4.954261954261955,
      "grad_norm": 0.00029671346419490874,
      "learning_rate": 1.0395010395010397e-07,
      "loss": 0.0,
      "step": 4766
    },
    {
      "epoch": 4.955301455301456,
      "grad_norm": 0.12956593930721283,
      "learning_rate": 1.0164010164010164e-07,
      "loss": 0.0037,
      "step": 4767
    },
    {
      "epoch": 4.956340956340957,
      "grad_norm": 6.041904449462891,
      "learning_rate": 9.933009933009933e-08,
      "loss": 0.1916,
      "step": 4768
    },
    {
      "epoch": 4.957380457380458,
      "grad_norm": 2.301347255706787,
      "learning_rate": 9.702009702009702e-08,
      "loss": 0.0397,
      "step": 4769
    },
    {
      "epoch": 4.958419958419959,
      "grad_norm": 11.286805152893066,
      "learning_rate": 9.471009471009472e-08,
      "loss": 0.6097,
      "step": 4770
    },
    {
      "epoch": 4.95945945945946,
      "grad_norm": 1.3557261228561401,
      "learning_rate": 9.240009240009241e-08,
      "loss": 0.0239,
      "step": 4771
    },
    {
      "epoch": 4.960498960498961,
      "grad_norm": 1.0138343572616577,
      "learning_rate": 9.00900900900901e-08,
      "loss": 0.0236,
      "step": 4772
    },
    {
      "epoch": 4.961538461538462,
      "grad_norm": 4.4559006690979,
      "learning_rate": 8.778008778008778e-08,
      "loss": 0.1399,
      "step": 4773
    },
    {
      "epoch": 4.962577962577963,
      "grad_norm": 5.926146984100342,
      "learning_rate": 8.547008547008549e-08,
      "loss": 0.3067,
      "step": 4774
    },
    {
      "epoch": 4.963617463617464,
      "grad_norm": 9.967759132385254,
      "learning_rate": 8.316008316008317e-08,
      "loss": 0.5148,
      "step": 4775
    },
    {
      "epoch": 4.964656964656965,
      "grad_norm": 1.4153497219085693,
      "learning_rate": 8.085008085008086e-08,
      "loss": 0.024,
      "step": 4776
    },
    {
      "epoch": 4.965696465696466,
      "grad_norm": 0.16361723840236664,
      "learning_rate": 7.854007854007855e-08,
      "loss": 0.0024,
      "step": 4777
    },
    {
      "epoch": 4.966735966735967,
      "grad_norm": 0.07358971983194351,
      "learning_rate": 7.623007623007624e-08,
      "loss": 0.0019,
      "step": 4778
    },
    {
      "epoch": 4.967775467775468,
      "grad_norm": 5.493592739105225,
      "learning_rate": 7.392007392007392e-08,
      "loss": 0.1629,
      "step": 4779
    },
    {
      "epoch": 4.968814968814969,
      "grad_norm": 6.8674139976501465,
      "learning_rate": 7.161007161007161e-08,
      "loss": 0.2934,
      "step": 4780
    },
    {
      "epoch": 4.96985446985447,
      "grad_norm": 3.8765203952789307,
      "learning_rate": 6.93000693000693e-08,
      "loss": 0.117,
      "step": 4781
    },
    {
      "epoch": 4.970893970893971,
      "grad_norm": 4.589378833770752,
      "learning_rate": 6.6990066990067e-08,
      "loss": 0.3599,
      "step": 4782
    },
    {
      "epoch": 4.971933471933472,
      "grad_norm": 0.12670215964317322,
      "learning_rate": 6.468006468006469e-08,
      "loss": 0.0035,
      "step": 4783
    },
    {
      "epoch": 4.972972972972973,
      "grad_norm": 6.151996612548828,
      "learning_rate": 6.237006237006238e-08,
      "loss": 0.2022,
      "step": 4784
    },
    {
      "epoch": 4.974012474012474,
      "grad_norm": 10.151179313659668,
      "learning_rate": 6.006006006006006e-08,
      "loss": 0.5903,
      "step": 4785
    },
    {
      "epoch": 4.975051975051975,
      "grad_norm": 3.344877004623413,
      "learning_rate": 5.775005775005776e-08,
      "loss": 0.0535,
      "step": 4786
    },
    {
      "epoch": 4.976091476091476,
      "grad_norm": 6.596832752227783,
      "learning_rate": 5.544005544005545e-08,
      "loss": 0.5492,
      "step": 4787
    },
    {
      "epoch": 4.977130977130977,
      "grad_norm": 2.9688796997070312,
      "learning_rate": 5.313005313005313e-08,
      "loss": 0.0638,
      "step": 4788
    },
    {
      "epoch": 4.978170478170478,
      "grad_norm": 1.099982500076294,
      "learning_rate": 5.082005082005082e-08,
      "loss": 0.0387,
      "step": 4789
    },
    {
      "epoch": 4.979209979209979,
      "grad_norm": 4.487499713897705,
      "learning_rate": 4.851004851004851e-08,
      "loss": 0.1387,
      "step": 4790
    },
    {
      "epoch": 4.98024948024948,
      "grad_norm": 8.446113586425781,
      "learning_rate": 4.6200046200046205e-08,
      "loss": 1.1905,
      "step": 4791
    },
    {
      "epoch": 4.981288981288982,
      "grad_norm": 8.844257354736328,
      "learning_rate": 4.389004389004389e-08,
      "loss": 0.4875,
      "step": 4792
    },
    {
      "epoch": 4.982328482328482,
      "grad_norm": 0.06120000034570694,
      "learning_rate": 4.158004158004159e-08,
      "loss": 0.0015,
      "step": 4793
    },
    {
      "epoch": 4.983367983367984,
      "grad_norm": 6.0494279861450195,
      "learning_rate": 3.9270039270039275e-08,
      "loss": 0.3082,
      "step": 4794
    },
    {
      "epoch": 4.984407484407484,
      "grad_norm": 9.200035095214844,
      "learning_rate": 3.696003696003696e-08,
      "loss": 0.4815,
      "step": 4795
    },
    {
      "epoch": 4.985446985446986,
      "grad_norm": 0.027429446578025818,
      "learning_rate": 3.465003465003465e-08,
      "loss": 0.0007,
      "step": 4796
    },
    {
      "epoch": 4.986486486486487,
      "grad_norm": 2.307527542114258,
      "learning_rate": 3.2340032340032345e-08,
      "loss": 0.0316,
      "step": 4797
    },
    {
      "epoch": 4.987525987525988,
      "grad_norm": 4.906373500823975,
      "learning_rate": 3.003003003003003e-08,
      "loss": 0.1492,
      "step": 4798
    },
    {
      "epoch": 4.988565488565489,
      "grad_norm": 1.1830755472183228,
      "learning_rate": 2.7720027720027723e-08,
      "loss": 0.0295,
      "step": 4799
    },
    {
      "epoch": 4.98960498960499,
      "grad_norm": 7.704885482788086,
      "learning_rate": 2.541002541002541e-08,
      "loss": 0.4183,
      "step": 4800
    },
    {
      "epoch": 4.990644490644491,
      "grad_norm": 1.4846118688583374,
      "learning_rate": 2.3100023100023102e-08,
      "loss": 0.0445,
      "step": 4801
    },
    {
      "epoch": 4.991683991683992,
      "grad_norm": 0.034488894045352936,
      "learning_rate": 2.0790020790020793e-08,
      "loss": 0.0008,
      "step": 4802
    },
    {
      "epoch": 4.992723492723493,
      "grad_norm": 1.4385161399841309,
      "learning_rate": 1.848001848001848e-08,
      "loss": 0.0268,
      "step": 4803
    },
    {
      "epoch": 4.993762993762994,
      "grad_norm": 10.849470138549805,
      "learning_rate": 1.6170016170016172e-08,
      "loss": 0.7787,
      "step": 4804
    },
    {
      "epoch": 4.994802494802495,
      "grad_norm": 0.08578181266784668,
      "learning_rate": 1.3860013860013862e-08,
      "loss": 0.0029,
      "step": 4805
    },
    {
      "epoch": 4.995841995841996,
      "grad_norm": 1.4152899980545044,
      "learning_rate": 1.1550011550011551e-08,
      "loss": 0.0174,
      "step": 4806
    },
    {
      "epoch": 4.996881496881497,
      "grad_norm": 8.894526481628418,
      "learning_rate": 9.24000924000924e-09,
      "loss": 0.4251,
      "step": 4807
    },
    {
      "epoch": 4.997920997920998,
      "grad_norm": 0.022568803280591965,
      "learning_rate": 6.930006930006931e-09,
      "loss": 0.0005,
      "step": 4808
    },
    {
      "epoch": 4.998960498960499,
      "grad_norm": 5.386369228363037,
      "learning_rate": 4.62000462000462e-09,
      "loss": 0.1613,
      "step": 4809
    },
    {
      "epoch": 5.0,
      "grad_norm": 5.23478364944458,
      "learning_rate": 2.31000231000231e-09,
      "loss": 0.1585,
      "step": 4810
    },
    {
      "epoch": 5.0,
      "step": 4810,
      "total_flos": 0.0,
      "train_loss": 0.37849533693159876,
      "train_runtime": 13899.499,
      "train_samples_per_second": 2.768,
      "train_steps_per_second": 0.346
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4810,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}