The response is too short to extract answer on GPQA. What should I set to extend it?

[URRealHero]

lm_eval --model local-chat-completions --tasks gpqa_main_cot_zeroshot --model_args model=Qwen/Qwen2-72B-Instruct,base_url=https://api.together.xyz/v1 --output_path ./gpqa/result/Qwen2 --use_cache ./gpqa/cache/Qwen2 --log_samples --limit 10 --gen_kwargs temperature=0.7,max_tokens=8192 Using this command, The Qwen2's result just end sooo weirdly like the image below

To be specific, only 256 tokens are generated. I'm wondering why this happens, is there any problem with max_tokens?