Training checkpoint 540000 - 5.5B tokens

README.md

@@ -18,11 +18,11 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 538,000
-- **Tokens Processed**: 5.51B tokens
-- **Current Loss**: 4.6078
+- **Training Step**: 540,000
+- **Tokens Processed**: 5.53B tokens
+- **Current Loss**: 4.6760
 - **Spike Rate**: 0.0508
-- **Learning Rate**: 8.30e-05
+- **Learning Rate**: 8.29e-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_538000.pt"
+    filename="checkpoint_540000.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**: 36.7% complete towards 15B tokens
+**Training Progress**: 36.9% complete towards 15B tokens

checkpoint_540000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12ed026f2216bef7a656ca1c11afd27860afe9d2ac86dad76941dd4485834d52
+size 999026730

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 538000,
-  "tokens_processed": 5509120000,
-  "loss": 4.607837925084207,
-  "spike_rate": 0.050761995090000456,
-  "learning_rate": 8.304732811135012e-05,
-  "epoch": 1.101824,
-  "progress_percent": 36.72746666666667
+  "training_step": 540000,
+  "tokens_processed": 5529600000,
+  "loss": 4.676038708841041,
+  "spike_rate": 0.050808739849472394,
+  "learning_rate": 8.286133714523995e-05,
+  "epoch": 1.10592,
+  "progress_percent": 36.864000000000004
 }