Training checkpoint 80000 - 0.8B tokens

README.md

@@ -18,9 +18,9 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 78,000
-- **Tokens Processed**: 0.80B tokens
-- **Current Loss**: 4.7427
+- **Training Step**: 80,000
+- **Tokens Processed**: 0.82B tokens
+- **Current Loss**: 4.6557
 - **Spike Rate**: 0.0502
 - **Learning Rate**: 1.99e-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_78000.pt"
+    filename="checkpoint_80000.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**: 5.3% complete towards 15B tokens
+**Training Progress**: 5.5% complete towards 15B tokens

checkpoint_80000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c3e05fa8d72f70574a52c8c948e61af7e124bca0567937dd88c1b446dab2e52
+size 999026492

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 78000,
-  "tokens_processed": 798720000,
-  "loss": 4.742708314575408,
-  "spike_rate": 0.050204712333369435,
-  "learning_rate": 0.0001994329398922472,
-  "epoch": 0.159744,
-  "progress_percent": 5.324813631522897
+  "training_step": 80000,
+  "tokens_processed": 819200000,
+  "loss": 4.655669050952178,
+  "spike_rate": 0.050184656109373844,
+  "learning_rate": 0.00019938546168400735,
+  "epoch": 0.16384,
+  "progress_percent": 5.461347314382458
 }