Training checkpoint 70000 - 0.7B tokens

README.md

@@ -18,9 +18,9 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 68,000
-- **Tokens Processed**: 0.70B tokens
-- **Current Loss**: 4.6944
+- **Training Step**: 70,000
+- **Tokens Processed**: 0.72B tokens
+- **Current Loss**: 4.6952
 - **Spike Rate**: 0.0502
 - **Learning Rate**: 2.00e-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_68000.pt"
+    filename="checkpoint_70000.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**: 4.6% complete towards 15B tokens
+**Training Progress**: 4.8% complete towards 15B tokens

checkpoint_70000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae0f2e87c8097896035b3df3dade5446afd9fbb75ef6f13d0d6cef53e28ab680
+size 999026492

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 68000,
-  "tokens_processed": 696320000,
-  "loss": 4.694437745368935,
-  "spike_rate": 0.05024811077175662,
-  "learning_rate": 0.0001996417645201202,
-  "epoch": 0.139264,
-  "progress_percent": 4.64214521722509
+  "training_step": 70000,
+  "tokens_processed": 716800000,
+  "loss": 4.695154943120782,
+  "spike_rate": 0.05023116208271157,
+  "learning_rate": 0.00019960381117747244,
+  "epoch": 0.14336,
+  "progress_percent": 4.778678900084651
 }