MCPcopy Create free account
hub / github.com/PaddlePaddle/FastDeploy / save_to_pytorch_benchmark_format

Function save_to_pytorch_benchmark_format

benchmarks/benchmark_serving.py:1013–1040  ·  view source on GitHub ↗

Save the benchmarking results to PyTorch Benchmark Format JSON file

(args: argparse.Namespace, results: dict[str, Any], file_name: str)

Source from the content-addressed store, hash-verified

1011
1012
1013def save_to_pytorch_benchmark_format(args: argparse.Namespace, results: dict[str, Any], file_name: str) -> None:
1014 """Save the benchmarking results to PyTorch Benchmark Format JSON file"""
1015 metrics = [
1016 "median_ttft_ms",
1017 "mean_ttft_ms",
1018 "std_ttft_ms",
1019 "p99_ttft_ms",
1020 "mean_tpot_ms",
1021 "median_tpot_ms",
1022 "std_tpot_ms",
1023 "p99_tpot_ms",
1024 "median_itl_ms",
1025 "mean_itl_ms",
1026 "std_itl_ms",
1027 "p99_itl_ms",
1028 ]
1029 # These raw data might be useful, but they are rather big. They can be added
1030 # later if needed
1031 ignored_metrics = ["ttfts", "itls", "generated_texts", "errors"]
1032 pt_records = convert_to_pytorch_benchmark_format(
1033 args=args,
1034 metrics={k: [results[k]] for k in metrics},
1035 extra_info={k: results[k] for k in results if k not in metrics and k not in ignored_metrics},
1036 )
1037 if pt_records:
1038 # Don't use json suffix here as we don't want CI to pick it up
1039 pt_file = f"{os.path.splitext(file_name)[0]}.pytorch.json"
1040 write_to_json(pt_file, pt_records)
1041
1042
1043def main(args: argparse.Namespace):

Callers 1

mainFunction · 0.70

Calls 2

write_to_jsonFunction · 0.90

Tested by

no test coverage detected