Training checkpoint 180000 - 1.8B tokens

README.md

@@ -18,9 +18,9 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 178,000
-- **Tokens Processed**: 1.82B tokens
-- **Current Loss**: 4.6877
+- **Training Step**: 180,000
+- **Tokens Processed**: 1.84B tokens
+- **Current Loss**: 4.6269
 - **Spike Rate**: 0.0504
 - **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_178000.pt"
+    filename="checkpoint_180000.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**: 12.2% complete towards 15B tokens
+**Training Progress**: 12.3% complete towards 15B tokens

checkpoint_180000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77d23433da60774c3d4a92b3a2935a82a117c6c237300c9449f4d26185a4614e
+size 999026666

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 178000,
-  "tokens_processed": 1822720000,
-  "loss": 4.687730422009585,
-  "spike_rate": 0.050382256686731526,
-  "learning_rate": 0.00019475548045602345,
-  "epoch": 0.364544,
-  "progress_percent": 12.15149777450097
+  "training_step": 180000,
+  "tokens_processed": 1843200000,
+  "loss": 4.626903585180932,
+  "spike_rate": 0.050403356740924936,
+  "learning_rate": 0.00019461474969843537,
+  "epoch": 0.36864,
+  "progress_percent": 12.28803145736053
 }