修改注释

src/train_model_grpo.py

@@ -119,7 +119,7 @@ class ModelTrainer:
             dtype=self.dtype,
             fast_inference = False, # Enable vLLM fast inference
             max_lora_rank = lora_rank,
-            gpu_memory_utilization=0.1,# Reduce if out of memory
+            gpu_memory_utilization=0.1, # 0.6 # Reduce if out of memory
         )
         
         # 添加 LoRA 适配器