Training checkpoint 554000 - 5.7B tokens

README.md

@@ -18,11 +18,11 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 542,000
-- **Tokens Processed**: 5.55B tokens
-- **Current Loss**: 4.6726
+- **Training Step**: 554,000
+- **Tokens Processed**: 5.67B tokens
+- **Current Loss**: 4.5783
 - **Spike Rate**: 0.0508
-- **Learning Rate**: 8.27e-05
+- **Learning Rate**: 8.15e-05
 
 ## 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_542000.pt"
+    filename="checkpoint_554000.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**: 37.0% complete towards 15B tokens
+**Training Progress**: 37.8% complete towards 15B tokens

checkpoint_554000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c78b28cb4c295b0562f1f1e3faa33b0843d95df41c1a0bfc36595ee445880d11
+size 999026730

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 542000,
-  "tokens_processed": 5550080000,
-  "loss": 4.672573742633761,
-  "spike_rate": 0.05079471937811068,
-  "learning_rate": 8.267454235145837e-05,
-  "epoch": 1.110016,
-  "progress_percent": 37.00053333333334
+  "training_step": 554000,
+  "tokens_processed": 5672960000,
+  "loss": 4.578331639736606,
+  "spike_rate": 0.05075958974589487,
+  "learning_rate": 8.15371506031409e-05,
+  "epoch": 1.134592,
+  "progress_percent": 37.81973333333333
 }