Training checkpoint 110000 - 1.1B tokens

README.md

@@ -18,11 +18,11 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 108,000
-- **Tokens Processed**: 1.11B tokens
-- **Current Loss**: 4.5420
+- **Training Step**: 110,000
+- **Tokens Processed**: 1.13B tokens
+- **Current Loss**: 4.6097
 - **Spike Rate**: 0.0503
-- **Learning Rate**: 1.99e-04
+- **Learning Rate**: 1.98e-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_108000.pt"
+    filename="checkpoint_110000.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**: 7.4% complete towards 15B tokens
+**Training Progress**: 7.5% complete towards 15B tokens

checkpoint_110000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2282effcd0d2e35ab08882e5e90edac2429365050fe3adff8d6c95014d9c586b
+size 999026666

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 108000,
-  "tokens_processed": 1105920000,
-  "loss": 4.54200165254624,
-  "spike_rate": 0.05027944500694136,
-  "learning_rate": 0.000198521495859823,
-  "epoch": 0.221184,
-  "progress_percent": 7.372818874416319
+  "training_step": 110000,
+  "tokens_processed": 1126400000,
+  "loss": 4.609737979561926,
+  "spike_rate": 0.05033084232855387,
+  "learning_rate": 0.00019844559766969382,
+  "epoch": 0.22528,
+  "progress_percent": 7.5093525572758795
 }