换用github jwjohns/unsloth-GRPO-qwen2.5 验证GRPO训练模型

src/qwen_notebook_clone.py

@@ -158,7 +158,7 @@ training_args = GRPOConfig(
     # Enable detailed metrics logging
     log_level = "info",
     disable_tqdm = False,  # Ensure progress bars are displayed
-    logging_steps = 5,  # Log metrics frequently
+    # logging_steps = 5,  # Log metrics frequently
     evaluation_strategy = "no",  # Disable evaluation since we don't have an eval dataset
 )