Training checkpoint 36000 - 0.4B tokens

README.md

@@ -18,10 +18,10 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 34,000
-- **Tokens Processed**: 0.35B tokens
-- **Current Loss**: 4.6213
-- **Spike Rate**: 0.0200
+- **Training Step**: 36,000
+- **Tokens Processed**: 0.37B tokens
+- **Current Loss**: 4.7309
+- **Spike Rate**: 0.0229
 - **Learning Rate**: 2.00e-04
 
 ## 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_34000.pt"
+    filename="checkpoint_36000.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**: 2.3% complete towards 15B tokens
+**Training Progress**: 2.5% complete towards 15B tokens

checkpoint_36000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bea74e7a701d737b0ad52bee47267e08557654ae4cfb1109035322e6c4c95c98
+size 999026492

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 34000,
-  "tokens_processed": 348160000,
-  "loss": 4.621290160612954,
-  "spike_rate": 0.019981226479303144,
-  "learning_rate": 0.00019999470285352774,
-  "epoch": 0.069632,
-  "progress_percent": 2.321072608612545
+  "training_step": 36000,
+  "tokens_processed": 368640000,
+  "loss": 4.730866090005601,
+  "spike_rate": 0.022853559834219567,
+  "learning_rate": 0.0001999892426494225,
+  "epoch": 0.073728,
+  "progress_percent": 2.457606291472106
 }