[Bug]: qwen3-1.7b output text truncate when serving with vllm docker。在默认参数下，vllm推理结果异常截断

[jjxyai]

Your current environment

Run with docker, system ubuntu 22.04, gpu mem 16G docker run --name vllm --gpus "device=1" -v ~/.cache/huggingface:/root/.cache/huggingface -v /home/user/mnt/:/vllm-workspace -w /vllm-workspace -p 8000:8000 --ipc=host vllm/vllm-openai:latest --config /vllm-workspace/weights/config.yaml ``` # config.yaml model: /vllm-workspace/weights/Qwen3-1.7B uvicorn-log-level: "info" host: "0.0.0.0" port: 8000

served-model-name: "Qwen3-1.7b"
max-model-len: 4096
max-num-seqs: 4
gpu-memory-utilization: 0.6
trust-remote-code: true
generation-config: "auto"
reasoning-parser: "qwen3"


```text

🐛 Describe the bug

output by vllm api:

while infer with huggingface transformers, result in:

'“张岱的《湖心亭看雪》”中，“大雪三日”后面一句是：\n\n**“天与云与山与水，上下一白。”**\n\n这句话出自张岱的《湖心亭看雪》，原文如下：\n\n> “余强饮三大白醉，间从秦氏口授童子文。中圣天竺国，见大雪三日，天与云与山与水，上下一白。”\n\n完整句子是：\n\n> “天与云与山与水，上下一白。”\n\n这是描写雪景非常传神的一句，表达了作者在雪夜赏雪的孤寂与清冷，同时也体现了文人的闲适心境。'

vllm api only return the first few words “张岱的《湖心亭看雪》”中，“大雪”，
are there any params to set？or something went wrong?

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[jjxyai]

also tried with

docker run --gpus all \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=$HF_TOKEN" \
    -p 8000:8000 \
    --ipc=host \
    vllm/vllm-openai:v0.12.0\
    --model Qwen/Qwen3-1.7B

got the same problem

[jjxyai]

newbie's problem.

solved by add "max_tokens": 1024 to the request body.

{ "model": "Qwen3-1.7B", "max_tokens": 1024, "prompt": ["张岱的湖心亭记中，"大雪三日"的后面一句是什么？"] }