mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-27 22:23:25 +08:00
4bb65f216f
* Update TensorRT-LLM --------- Co-authored-by: meghagarwal <16129366+megha95@users.noreply.github.com> Co-authored-by: Shixiaowei02 <39303645+Shixiaowei02@users.noreply.github.com>
19 lines
424 B
Bash
19 lines
424 B
Bash
#!/bin/bash
|
|
set -ex
|
|
|
|
PROMPT="Tell a story"
|
|
LLAMA_MODEL_DIR=$1
|
|
WORLD_SIZE=${2:-2}
|
|
|
|
dir=$(dirname "$0")
|
|
|
|
python3 $dir/llm_examples.py --task run_llm_with_auto_parallel \
|
|
--prompt="$PROMPT" \
|
|
--world_size=$WORLD_SIZE \
|
|
--hf_model_dir=$LLAMA_MODEL_DIR
|
|
|
|
python3 $dir/llm_examples.py --task run_llm_with_auto_parallel_async \
|
|
--prompt="$PROMPT" \
|
|
--world_size=$WORLD_SIZE \
|
|
--hf_model_dir=$LLAMA_MODEL_DIR
|