Skip to content

请问对于ovis模型,怎么正确将MAX_PARTITION传入?依靠环境变量vllm serve启动不生效 #4881

Open
@AstonyJ

Description

@AstonyJ

export NPROC_PER_NODE=1
export CUDA_VISIBLE_DEVICES=1
export MAX_PARTITION=4

vllm serve ${MODEL_PATH} --host 0.0.0.0 --tensor-parallel-size 1 --trust-remote-code --port 8056 --max-model-len 32768 --gpu-memory-utilization 0.9 --max-num-batched-tokens 8192

请问对于ovis模型,怎么正确将MAX_PARTITION传入?依靠环境变量vllm serve启动不生效

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions