20231088/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[Misc][Benchmark] Add support for different tokenizer_mode (#15040)

Browse Source 
Signed-off-by: Aaron Pham <contact@aarnphm.xyz>
Signed-off-by: Russell Bryant <rbryant@redhat.com>
Co-authored-by: Russell Bryant <rbryant@redhat.com>
This commit is contained in:
Aaron Pham 2025-03-19 10:56:50 -04:00 committed by GitHub
parent 073d1ed354
commit 6c5a3195db
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 13 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
benchmarks/benchmark_serving_structured_output.py
View File

@ -732,8 +732,11 @@ def main(args: argparse.Namespace):
api_url = f"http://{args.host}:{args.port}{args.endpoint}" api_url = f"http://{args.host}:{args.port}{args.endpoint}"
base_url = f"http://{args.host}:{args.port}" base_url = f"http://{args.host}:{args.port}"
tokenizer = get_tokenizer(tokenizer_id, tokenizer = get_tokenizer(
trust_remote_code=args.trust_remote_code) tokenizer_id,
trust_remote_code=args.trust_remote_code,
tokenizer_mode=args.tokenizer_mode,
)
if args.dataset == 'grammar': if args.dataset == 'grammar':
args.structure_type = 'guided_grammar' args.structure_type = 'guided_grammar'
@ -876,6 +879,13 @@ if __name__ == "__main__":
help= help=
"Name or path of the tokenizer, if not using the default tokenizer.", # noqa: E501 "Name or path of the tokenizer, if not using the default tokenizer.", # noqa: E501
) )
parser.add_argument(
"--tokenizer-mode",
type=str,
default="auto",
help=
"Name or path of the tokenizer, if not using the default tokenizer.", # noqa: E501
)
parser.add_argument( parser.add_argument(
"--num-prompts", "--num-prompts",
type=int, type=int,

1
benchmarks/run_structured_output_benchmark.sh
View File

@ -54,6 +54,7 @@ for qps in "${QPS_VALUES[@]}"; do
python "$SCRIPT_DIR/benchmark_serving_structured_output.py" $COMMON_PARAMS \ python "$SCRIPT_DIR/benchmark_serving_structured_output.py" $COMMON_PARAMS \
--request-rate $qps \ --request-rate $qps \
--result-filename "$FILENAME" \ --result-filename "$FILENAME" \
--tokenizer-mode ${TOKENIZER_MODE:-"auto"} \
--port ${PORT:-8000} --port ${PORT:-8000}
echo "Completed benchmark with QPS: $qps" echo "Completed benchmark with QPS: $qps"