Internal Server Error 500 in VLLM while using some images.

#12
by as-sriram - opened

Hosting the model via command:
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m vllm.entrypoints.openai.api_server --model /models/Qwen/Qwen2.5-VL-32B-Instruct --dtype auto --tensor-parallel-size 4 --served-model-name qwen_2.5_32b_vl --max_model_len 128000 --tokenizer_mode auto --limit-mm-per-prompt image\=10 --disable-log-requests

Here are the sample list of images which are leading to the error. Can someone please help me with the issue. Thanks.
3037e71f7ad74ebcbf66f67e647e2fe2.png
ef40b5f59e76492bb1b5460f5cd69c71.png
d97b85b2acf54fe49cb82f705523b334.png
d1994f7e94e243829d112df7b31b7647.png

as-sriram changed discussion title from Internal Server error 500 in VLLM while using some images. to Internal Server Error 500 in VLLM while using some images.

got the same problem there

Sign up or log in to comment