dyhuachi

> > vllm启动代码：CUDA_VISIBLE_DEVICES=0,1,2,3 vllm serve /mnt/disk1/DATA/llm/Qwen3-VL-32B-Instr Qwen3-VL-32B-Instruct --dtype float16 --tensor-parallel-size 4 --gpu-memory-utilization 0.8 --port 9996 --max-model-ling --media-io-kwargs '{"video": {"num_frames": -1}}' 调用代码：import base64 import numpy as np from PIL import Image...