Upload sft_03_train.py with huggingface_hub

sft_03_train.py

@@ -184,6 +184,7 @@ def sample_test(model, tokenizer, prompt_text: str, max_new: int = 200):
 
 # ===== Main =====
 def main():
+    global BATCH_SIZE, GRAD_ACCUM_STEPS
     parser = argparse.ArgumentParser()
     parser.add_argument("--base", type=str,
                         default="runs/tr-200m-v5/best_ckpt.pt",
@@ -205,7 +206,6 @@ def main():
                         default="max-autotune-no-cudagraphs")
     args = parser.parse_args()
 
-    global BATCH_SIZE, GRAD_ACCUM_STEPS
     BATCH_SIZE = args.batch
     GRAD_ACCUM_STEPS = args.grad_accum