mem 用完自动kill

[missTL]

Reminder

• I have read the above rules and searched the existing issues.

System Info

• llamafactory version: 0.9.2.dev0
• Platform: Linux-5.4.225-1.el7.elrepo.x86_64-x86_64-with-glibc2.17
• Python version: 3.10.16
• PyTorch version: 2.5.1+cu124 (GPU)
• Transformers version: 4.49.0
• Datasets version: 3.1.0
• Accelerate version: 1.0.1
• PEFT version: 0.12.0
• TRL version: 0.9.6
• GPU type: NVIDIA RTX A6000
• DeepSpeed version: 0.14.4
• Bitsandbytes version: 0.45.0
• vLLM version: 0.7.3

Reproduction

mem会用完，并自动kill。500G内存都不够吗？val_128world_motion_muti数据只有5000条样本
VLLM推理命令：CUDA_VISIBLE_DEVICES='4,5,6,7' python scripts/vllm_infer.py --model_name_or_path  /home/zengshuang.zs/output/llm/v4.3_128 --dataset val_128world_motion_muti --template qwen2_vl --cutoff_len 32768 --max_new_tokens 3500 --max_samples 100000 --image_resolution 524288 --save_name world_v4.3_128.jsonl --temperature 0.1 --top_p 0.1 --top_k 10

Others

No response