Training checkpoint 172000 - 1.8B tokens

README.md

@@ -18,9 +18,9 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 170,000
-- **Tokens Processed**: 1.74B tokens
-- **Current Loss**: 4.6092
+- **Training Step**: 172,000
+- **Tokens Processed**: 1.76B tokens
+- **Current Loss**: 4.6496
 - **Spike Rate**: 0.0503
 - **Learning Rate**: 1.95e-04
 
@@ -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_170000.pt"
+    filename="checkpoint_172000.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**: 11.6% complete towards 15B tokens
+**Training Progress**: 11.7% complete towards 15B tokens

checkpoint_172000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1853d16bc60ae7352a1917e950dc0c623f704c592bd08c4779fc335690d47b5f
+size 999026666

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 170000,
-  "tokens_processed": 1740800000,
-  "loss": 4.6092252222593535,
-  "spike_rate": 0.05033774667762338,
-  "learning_rate": 0.0001953002401076132,
-  "epoch": 0.34816,
-  "progress_percent": 11.605363043062724
+  "training_step": 172000,
+  "tokens_processed": 1761280000,
+  "loss": 4.649643607607375,
+  "spike_rate": 0.050314416365587575,
+  "learning_rate": 0.00019516677995436456,
+  "epoch": 0.352256,
+  "progress_percent": 11.741896725922285
 }