bench: add dataset param for bench_multiturn

[zeroorhero]

In bench_multiturn.py file, the dataset is downloaded from HuggingFace by default, which may sometimes fail. Therefore, a new parameter is added to specify the path to the local dataset.

Motivation

Modifications

Checklist

• Format your code according to the Code Formatting with Pre-Commit.
• Add unit tests as outlined in the Running Unit Tests.
• Update documentation / docstrings / example tutorials as needed, according to Writing Documentation.
• Provide throughput / latency benchmark results and accuracy evaluation results as needed, according to Benchmark and Profiling and Accuracy Results.
• For reviewers: If you haven't made any contributions to this PR and are only assisting with merging the main branch, please remove yourself as a co-author when merging the PR.
• Please feel free to join our Slack channel at https://slack.sglang.ai to discuss your PR.

[zeroorhero]

cc @xiezhq-hermann

[xiezhq-hermann]

hi @zeroorhero thank you for your interest in using the benchmark. But I do not think it is a necessary nor a very good change, as if the default option is set to a local file it is even easier to fail for users without it in hand.

[zeroorhero]

hi @zeroorhero thank you for your interest in using the benchmark. But I do not think it is a necessary nor a very good change, as if the default option is set to a local file it is even easier to fail for users without it in hand.

In bench_serving.py, if the default path does not exist, it will proceed to download the dataset. Perhaps setting the default value to an empty string ("") is a good choice, and I will make this change later. However, in China, this option is still very much needed because it is often difficult to successfully download from Hugging Face.