Training checkpoint 140000 - 1.4B tokens

README.md

@@ -18,10 +18,10 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 138,000
-- **Tokens Processed**: 1.41B tokens
-- **Current Loss**: 4.6923
-- **Spike Rate**: 0.0502
+- **Training Step**: 140,000
+- **Tokens Processed**: 1.43B tokens
+- **Current Loss**: 4.6054
+- **Spike Rate**: 0.0503
 - **Learning Rate**: 1.97e-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_138000.pt"
+    filename="checkpoint_140000.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**: 9.4% complete towards 15B tokens
+**Training Progress**: 9.6% complete towards 15B tokens

checkpoint_140000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61260e656cce2e2f57f83f7c31984d5d129ca06add66ed5cfba7703101bcb5e8
+size 999026666

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 138000,
-  "tokens_processed": 1413120000,
-  "loss": 4.692317863431244,
-  "spike_rate": 0.05022373977899401,
-  "learning_rate": 0.0001971858896642582,
-  "epoch": 0.282624,
-  "progress_percent": 9.42082411730974
+  "training_step": 140000,
+  "tokens_processed": 1433600000,
+  "loss": 4.6054020905118,
+  "spike_rate": 0.050284036661693426,
+  "learning_rate": 0.00019708189852039206,
+  "epoch": 0.28672,
+  "progress_percent": 9.557357800169301
 }