Training checkpoint 64000 - 0.7B tokens

README.md

@@ -18,9 +18,9 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 62,000
-- **Tokens Processed**: 0.63B tokens
-- **Current Loss**: 4.5982
+- **Training Step**: 64,000
+- **Tokens Processed**: 0.66B tokens
+- **Current Loss**: 4.7044
 - **Spike Rate**: 0.0502
 - **Learning Rate**: 2.00e-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_62000.pt"
+    filename="checkpoint_64000.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**: 4.2% complete towards 15B tokens
+**Training Progress**: 4.4% complete towards 15B tokens

checkpoint_64000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb22d6542c61b1d18c0d31260aab5b4130ee1e07e50f0ea921e5b845d0d41db2
+size 999026492

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 62000,
-  "tokens_processed": 634880000,
-  "loss": 4.598159832880502,
-  "spike_rate": 0.05019693164700216,
-  "learning_rate": 0.00019974417802019376,
-  "epoch": 0.126976,
-  "progress_percent": 4.232544168646405
+  "training_step": 64000,
+  "tokens_processed": 655360000,
+  "loss": 4.7044141859149216,
+  "spike_rate": 0.050169520119615126,
+  "learning_rate": 0.00019971194861373765,
+  "epoch": 0.131072,
+  "progress_percent": 4.369077851505967
 }