mlc-ai/xgrammar v0.1.34 on GitHub

What's Changed

Reapply "refactor: migrate the binding logic into tvm_ffi. (#550)"" by @Seven-Streams in #576
fix: accept {n, -1} as unbounded repeat in EBNF parser by @ushiromiya-lion in #579
fix: AnyTokensFormat with exclude_tokens should be treated as self-terminating by @ushiromiya-lion in #578
fix: fix the building of the website. by @Seven-Streams in #582
doc: update the citation part of the readme. by @Seven-Streams in #583
refac: remove the restriction of unlimited. by @Seven-Streams in #585
feat: add builtin structural tag support for Gemma 4 models by @leuski in #588
fix: auto-skip HuggingFace tests when HF_TOKEN is not set by @ianliuy in #589
fix the broken link of api/python/builtin_structural_tag. by @Seven-Streams in #592
docs: rewrite structural tag tutorial with clearer structure by @Ubospica in #590
chore: disable the unit test with python3.14t and macOS-14. by @Seven-Streams in #598
feat: support tool_choice for get_builtin_structural_tag. by @Seven-Streams in #586
fix structural tag validation and rename builtin tests by @Ubospica in #601
fix: fix the type compatibility. by @Seven-Streams in #602
chore: disable the use of hf_token when not hf_token_required is explicitly passed. by @Seven-Streams in #604
feat: support DeepSeek_v4's tool calling format. by @Seven-Streams in #596
fix: fix the DeepSeek_R1's built-in tool-calling format. by @Seven-Streams in #597
feat: support Qwen3.6's tool-calling format. by @Seven-Streams in #603
Bump to v0.1.34. by @Seven-Streams in #605

Full Changelog: v0.1.33...v0.1.34