Training checkpoint 164000 - 1.7B tokens

README.md

@@ -18,10 +18,10 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 162,000
-- **Tokens Processed**: 1.66B tokens
-- **Current Loss**: 4.6291
-- **Spike Rate**: 0.0503
+- **Training Step**: 164,000
+- **Tokens Processed**: 1.68B tokens
+- **Current Loss**: 4.6331
+- **Spike Rate**: 0.0504
 - **Learning Rate**: 1.96e-04
 
 ## Model Architecture
@@ -54,7 +54,7 @@ This represents ongoing training of the first large-scale spiking neural network
 from huggingface_hub import hf_hub_download
 checkpoint = hf_hub_download(
     repo_id="rootxhacker/piking-llm-5b-3epochs-exp",
-    filename="checkpoint_162000.pt"
+    filename="checkpoint_164000.pt"
 )
 
 # Load with custom spiking model code
@@ -65,4 +65,4 @@ checkpoint = hf_hub_download(
 
 **🔬 This is live research in progress! Check back for updates as training continues.**
 
-**Training Progress**: 11.1% complete towards 15B tokens
+**Training Progress**: 11.2% complete towards 15B tokens

checkpoint_164000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:818b78166ba8774809ad6ef93e8ce5808998bee0ad32f387b84d2400914eba90
+size 999026666

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 162000,
-  "tokens_processed": 1658880000,
-  "loss": 4.629077056898565,
-  "spike_rate": 0.05031083954575412,
-  "learning_rate": 0.00019581581437828552,
-  "epoch": 0.331776,
-  "progress_percent": 11.059228311624478
+  "training_step": 164000,
+  "tokens_processed": 1679360000,
+  "loss": 4.63306390764847,
+  "spike_rate": 0.050395141075132543,
+  "learning_rate": 0.00019568966573544215,
+  "epoch": 0.335872,
+  "progress_percent": 11.19576199448404
 }