kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-01-13 22:18:36 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

[None][feat] Add export data to build and run script for AD (#10299)

Browse Source 
Signed-off-by: Tal Cherckez <127761168+tcherckez-nvidia@users.noreply.github.com>
This commit is contained in:
tcherckez-nvidia 2026-01-01 11:54:47 +02:00 committed by GitHub
parent 9f5b750a93
commit 4868772ad7
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 6 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
examples/auto_deploy/build_and_run_ad.py
View File

@ -277,7 +277,11 @@ def main(config: Optional[ExperimentConfig] = None):
config.prompt.queries,
sampling_params=SamplingParams(**config.prompt.sp_kwargs),
)
results = {"prompts_and_outputs": print_outputs(outs)}
results = {
"prompts_and_outputs": print_outputs(outs),
}
# Add config values so they get logged to JET extra
results.update(config.model_dump(mode="json"))
# run a benchmark for the model with batch_size == config.benchmark_bs
if config.benchmark.enabled and config.args.runtime != "trtllm":

2
tensorrt_llm/_torch/auto_deploy/utils/benchmark.py
View File

@ -122,4 +122,4 @@ def store_benchmark_results(results: dict, results_path: str):
results_path = pathlib.Path(results_path)
results_path.parent.mkdir(parents=True, exist_ok=True)
with results_path.open("w") as results_file:
json.dump(results, results_file)
json.dump(results, results_file, indent=2)