Hugging Face's logo

openbmb
/
MiniCPM-V-4_5

Image-Text-to-Text
Transformers
Safetensors
multilingual
minicpmv
feature-extraction
minicpm-v
vision
ocr
multi-image
video
custom_code
conversational
Model card Files
xet
Community
20

Enable thinking not works as expected while using VLLM

#16
by MRU4913 - opened
Discussion
MRU4913
•
edited Sep 16, 2025

{
"bos_token_id": 151643,
"do_sample": true,
"eos_token_id": [
151645,
151643
],
"pad_token_id": 151643,
"temperature": 0.6,
"top_k": 20,
"top_p": 0.95,
"chat_template_kwargs": {"enable_thinking": false},
"transformers_version": "4.51.0"
}

I'm using vlllm for deployment, but it still outputs <think>
tc-mb
 OpenBMB org

https://github.com/OpenSQZ/MiniCPM-V-CookBook/blob/main/deployment/vllm/minicpm-v4_5_vllm.md
Maybe this cookbook can give you some reference.

MRU4913 changed discussion status to closed

Sign up or log in to comment