Training checkpoint 120000 - 1.2B tokens

README.md

@@ -18,9 +18,9 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 118,000
-- **Tokens Processed**: 1.21B tokens
-- **Current Loss**: 4.5599
+- **Training Step**: 120,000
+- **Tokens Processed**: 1.23B tokens
+- **Current Loss**: 4.4975
 - **Spike Rate**: 0.0504
 - **Learning Rate**: 1.98e-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_118000.pt"
+    filename="checkpoint_120000.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**: 8.1% complete towards 15B tokens
+**Training Progress**: 8.2% complete towards 15B tokens

checkpoint_120000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:beabac35a669bdc737834cbd0d16d5d2c79d49f66d97a536400a890aa251a06f
+size 999026666

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 118000,
-  "tokens_processed": 1208320000,
-  "loss": 4.559912255303242,
-  "spike_rate": 0.05038838251614075,
-  "learning_rate": 0.00019812317613561824,
-  "epoch": 0.241664,
-  "progress_percent": 8.055487288714126
+  "training_step": 120000,
+  "tokens_processed": 1228800000,
+  "loss": 4.49753803556334,
+  "spike_rate": 0.05036709069291224,
+  "learning_rate": 0.00019803787127949346,
+  "epoch": 0.24576,
+  "progress_percent": 8.192020971573687
 }