opt-babylm2-rewritten-clean-spacy_no-num-adj-earlystop-strictest-bpe_seed-42_1e-3

This model was trained from scratch on the kanishka/babylm2-rewritten-clean-spacy_no-num-adj-strictest dataset. It achieves the following results on the evaluation set:

• Loss: 2.6448
• Accuracy: 0.4821

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.001
• train_batch_size: 32
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 32000
• num_epochs: 20.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.0903	1.0	17737	3.1599	0.4223
2.9711	2.0	35474	3.0540	0.4325
2.8966	3.0	53211	2.9907	0.4393
2.8662	4.0	70948	2.9624	0.4425
2.8317	5.0	88685	2.9386	0.4454
2.8078	6.0	106422	2.9182	0.4469
2.7922	7.0	124159	2.9017	0.4491
2.7802	8.0	141896	2.8859	0.4509
2.7585	9.0	159633	2.8701	0.4525
2.7413	10.0	177370	2.8554	0.4540
2.7208	11.0	195107	2.8396	0.4561
2.6982	12.0	212844	2.8205	0.4578
2.673	13.0	230581	2.8026	0.4606
2.6515	14.0	248318	2.7836	0.4627
2.6193	15.0	266055	2.7617	0.4654
2.5807	16.0	283792	2.7413	0.4680
2.5416	17.0	301529	2.7143	0.4716
2.4836	18.0	319266	2.6875	0.4752
2.418	19.0	337003	2.6591	0.4792
2.3316	20.0	354740	2.6448	0.4821

Framework versions

• Transformers 4.48.0
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.1