1#!/bin/bash 2#SBATCH -A <account> 3#SBATCH -p <partition> 4#SBATCH -t 01:00:00 5#SBATCH -N 2 6#SBATCH --ntasks-per-node=8 7#SBATCH -o logs/trtllm-serve.out 8#SBATCH -e logs/trtllm-serve.err 9#SBATCH -J trtllm-serve101112# NOTE, this feature is experimental and may not work on all systems.13# The trtllm-llmapi-launch is a script that launches the LLM-API code on14# Slurm-like systems, and can support multi-node and multi-GPU setups.1516# Note that, the number of MPI processes should be the same as the model world17# size. e.g. For tensor_parallel_size=16, you may use 2 nodes with 8 gpus for18# each, or 4 nodes with 4 gpus for each or other combinations.1920# This docker image should have tensorrt_llm installed, or you need to install21# it in the task.2223# The following variables are expected to be set in the environment:24# You can set them via --export in the srun/sbatch command.25#   CONTAINER_IMAGE: the docker image to use, you'd better install tensorrt_llm in it, or install it in the task.26#   MOUNT_DIR: the directory to mount in the container27#   MOUNT_DEST: the destination directory in the container28#   WORKDIR: the working directory in the container29#   SOURCE_ROOT: the path to the TensorRT LLM source30#   PROLOGUE: the prologue to run before the script31#   LOCAL_MODEL: the local model directory to use, NOTE: downloading from HF is32#      not supported in Slurm mode, you need to download the model and put it in33#      the LOCAL_MODEL directory.3435echo"Starting trtllm-serve..."36# Just launch trtllm-serve job with trtllm-llmapi-launch command.37srun-l\38--container-image=${CONTAINER_IMAGE}\39--container-mounts=${MOUNT_DIR}:${MOUNT_DEST}\40--container-workdir=${WORKDIR}\41--export=ALL,PYTHONPATH=${SOURCE_ROOT}\42--mpi=pmix\43bash-c"44        set -ex45$PROLOGUE46        export PATH=$PATH:~/.local/bin4748        trtllm-llmapi-launch \49         trtllm-serve$LOCAL_MODEL \50            --tp_size 16 \51            --host 0.0.0.0 \52${ADDITIONAL_OPTIONS}53    "