Training checkpoint 122000 - 1.2B tokens

README.md

@@ -18,10 +18,10 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 120,000
-- **Tokens Processed**: 1.23B tokens
-- **Current Loss**: 4.4975
-- **Spike Rate**: 0.0504
+- **Training Step**: 122,000
+- **Tokens Processed**: 1.25B tokens
+- **Current Loss**: 4.5518
+- **Spike Rate**: 0.0503
 - **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_120000.pt"
+    filename="checkpoint_122000.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**: 8.2% complete towards 15B tokens
+**Training Progress**: 8.3% complete towards 15B tokens

checkpoint_122000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d5a8b0d59ead3979fbfb4f07d667033a9125ee626c6c3b1c036f2692ebac166
+size 999026666

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 120000,
-  "tokens_processed": 1228800000,
-  "loss": 4.49753803556334,
-  "spike_rate": 0.05036709069291224,
-  "learning_rate": 0.00019803787127949346,
-  "epoch": 0.24576,
-  "progress_percent": 8.192020971573687
+  "training_step": 122000,
+  "tokens_processed": 1249280000,
+  "loss": 4.551811754550579,
+  "spike_rate": 0.050348762208124606,
+  "learning_rate": 0.00019795068984869974,
+  "epoch": 0.249856,
+  "progress_percent": 8.328554654433248
 }