Training checkpoint 14000 - 0.1B tokens

README.md

@@ -18,11 +18,11 @@ pipeline_tag: text-generation
 
 ## Current Training Status
 
-- **Training Step**: 12,000
-- **Tokens Processed**: 0.12B tokens
-- **Current Loss**: 4.7274
-- **Spike Rate**: 0.0116
-- **Learning Rate**: 7.71e-05
+- **Training Step**: 14,000
+- **Tokens Processed**: 0.14B tokens
+- **Current Loss**: 4.7199
+- **Spike Rate**: 0.0117
+- **Learning Rate**: 9.73e-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_12000.pt"
+    filename="checkpoint_14000.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**: 0.8% complete towards 15B tokens
+**Training Progress**: 1.0% complete towards 15B tokens

checkpoint_14000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9b6375064d5834b44838d2c56ec1cd5ea64ca2f6237ae7729c2881042a5470d
+size 999026492

config.json

@@ -4,11 +4,11 @@
   "hidden_size": 768,
   "num_layers": 12,
   "max_seq_length": 1024,
-  "training_step": 12000,
-  "tokens_processed": 122880000,
-  "loss": 4.727390102970982,
-  "spike_rate": 0.011585013045629368,
-  "learning_rate": 7.710942751494316e-05,
-  "epoch": 0.024576,
-  "progress_percent": 0.8192020971573687
+  "training_step": 14000,
+  "tokens_processed": 143360000,
+  "loss": 4.719939580122661,
+  "spike_rate": 0.011710215059494818,
+  "learning_rate": 9.73403054507967e-05,
+  "epoch": 0.028672,
+  "progress_percent": 0.9557357800169302
 }