mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-02-06 19:21:52 +08:00
9bd15f1937
* TensorRT-LLM Release 0.10.0 --------- Co-authored-by: Loki <lokravi@amazon.com> Co-authored-by: meghagarwal <16129366+megha95@users.noreply.github.com>
31 lines
808 B
Python
31 lines
808 B
Python
#!/usr/bin/env python3
|
|
import os
|
|
|
|
import click
|
|
|
|
from tensorrt_llm.hlapi import LLM, ModelConfig, SamplingConfig
|
|
|
|
|
|
@click.command()
|
|
@click.option("--model_dir", type=str, required=True)
|
|
@click.option("--tp_size", type=int, required=True)
|
|
@click.option("--engine_dir", type=str, default=None)
|
|
def main(model_dir: str, tp_size: int, engine_dir: str):
|
|
config = ModelConfig(model_dir)
|
|
config.parallel_config.tp_size = tp_size
|
|
|
|
llm = LLM(config)
|
|
|
|
if engine_dir is not None and os.path.abspath(
|
|
engine_dir) != os.path.abspath(model_dir):
|
|
llm.save(engine_dir)
|
|
|
|
prompt = [45, 12, 13]
|
|
sampling_config = SamplingConfig(max_new_tokens=10)
|
|
for output in llm.generate([prompt], sampling_config=sampling_config):
|
|
print(output)
|
|
|
|
|
|
if __name__ == '__main__':
|
|
main()
|