Upload train_qwen3_codeforces.py with huggingface_hub

train_qwen3_codeforces.py

@@ -74,8 +74,8 @@ training_args = SFTConfig(
     bf16=True,
     max_grad_norm=1.0,
 
-    # Dataset formatting
-    dataset_text_field="",  # We use messages format
+    # Dataset formatting - use messages format
+    dataset_text_field="messages",  # Use messages format for chat
     packing=False,
 )