2.6. vllm run-batch arguments
Run batch inference jobs for the specified model.
$ vllm run-batch
- --disable-log-requests
Disable logging requests.
Default: False
- --disable-log-stats
Disable logging statistics.
Default: False
- --enable-metrics
Enables Prometheus metrics.
Default: False
- --enable-prompt-tokens-details
Enables
prompt_tokens_detailsin usage when set to True.Default: False
- --max-log-len MAX_LOG_LEN
Maximum number of prompt characters or prompt ID numbers printed in the log.
Default: Unlimited
- --output-tmp-dir OUTPUT_TMP_DIR
The directory to store the output file before uploading it to the output URL.
Default: None
- --port PORT
Port number for the Prometheus metrics server. Only needed if
enable-metricsis set.Default: 8000
- --response-role RESPONSE_ROLE
The role name to return if
request.add_generation_prompt=True.Default: assistant
- --url URL
Prometheus metrics server URL. Only required if
enable-metricsis set).Default: 0.0.0.0
- --use-v2-block-manager
DEPRECATED. Block manager v1 has been removed.
SelfAttnBlockSpaceManager(block manager v2) is now the default. Setting--use-v2-block-managerflag to True or False has no effect on vLLM behavior.Default: True
- -i INPUT_FILE, --input-file INPUT_FILE
The path or URL to a single input file. Supports local file paths and HTTP or HTTPS. If a URL is specified, the file should be available using HTTP GET.
Default: None
- -o OUTPUT_FILE, --output-file OUTPUT_FILE
The path or URL to a single output file. Supports local file paths and HTTP or HTTPS. If a URL is specified, the file should be available using HTTP PUT.
Default: None