xgrammar 0.1.13

[qeternity]

Motivation

Xgrammar compiler now supports regex compilation directly and the latest contains fixes (e.g. lazy groups) which bring it closer to parity with Outlines.

Modifications

Upgrade deps and migrate to regex compiler.

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.

[zhyncs]

There are some issues with 0.1.13, please wait for @Ubospica to release a new version.

[qeternity]

@zhyncs understood, thanks for the review and info.

I presume the issues you reference are build related? Installing from wheel passed our internal smoke tests fwiw.

[Ubospica]

@qeternity Yes the CI fails because xgrammar's new version contains a kernel not compatible with CUDA<12.4. It should work properly for CUDA>=12.4. We have fixed this issue (mlc-ai/xgrammar#206) and will release a new version later.

[qeternity]

@Ubospica gotcha. We are running 12.4 so I guess that explains things :)

[zhyncs]

Ah Currently SGLang is using CUDA 12.1 for the CIs :)
ref https://github.com/sgl-project/sglang/blob/main/docs/developer/setup_github_runner.md#step-1-start-a-docker-container
Maybe we should upgrade CUDA 12.4 @merrymercy @Ying1123