CUDA_VISIBLE_DEVICES=1,3,5,7 bash /mnt/volumes/base-ov-ali-sh-mix/yjl3/python_project/lmeval_project/eval_script/lmeval_to_eval.sh \ --model_path /lpai/inputs/models/Qwen__Qwen3_5-4B/26-05-25-0949 \ --model_name Qwen__Qwen3_5-4B \ --python /mnt/volumes/base-ov-ali-sh-mix/yjl3/miniforge3/envs/lmeval_vllm2/bin/python \ --tasks bbh \ --data_parallel_size 4 \ --apply_chat_template 1 \ --enable_thinking 0 \ --system_instruction "You are a helpful assistant." \ --think_end_token "</think>" \ --batch_size 32 \ --model_be vllm \ --gen_kwargs "do_sample=True,temperature=0.7,top_p=0.8,top_k=20,min_p=0.0,presence_penalty=1.5,repetition_penalty=1.0,max_gen_toks=4096" \ --gpu_memory_utilization 0.8

CUDA_VISIBLE_DEVICES=1,3,5,7 bash /mnt/volumes/base-ov-ali-sh-mix/yjl3/python_project/lmeval_project/eval_script/lmeval_to_eval.sh \ --model_path /lpai/inputs/models/Qwen__Qwen3_5-4B/26-05-25-0949 \ --model_name Qwen__Qwen3_5-4B \ --python /mnt/volumes/base-ov-ali-sh-mix/yjl3/miniforge3/envs/lmeval_vllm2/bin/python \ --tasks bbh \ --data_parallel_size 4 \ --apply_chat_template 1 \ --enable_thinking 0 \ --system_instruction "You are a helpful assistant." \ --think_end_token "</think>" \ --batch_size 32 \ --model_be vllm \ --gen_kwargs "do_sample=True,temperature=0.7,top_p=0.8,top_k=20,min_p=0.0,presence_penalty=1.5,repetition_penalty=1.0,max_gen_toks=4096" \ --gpu_memory_utilization 0.8