Motivation

I noticed that sglang stopped logging info level logs whenever I use multimodal models. E.g. there is no "the server is fired up and ready to roll" when I load any multimodal model:

I debugged a little bit and found out it's introduced in #5350: sglang loads all processors when loading multimodal model, and internvl introduced numpy.distutils which has a known bug (see: numpy/numpy#24213).

Upon further reading, I found out the introduction of this package might have been a bug itself, as the only location where cpu is used is actually expecting decord.cpu instead of numpy.distutils.cpuinfo.cpu, so I removed the line and the logging is working again:

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.