Liger Kernel

license

BSD-2-Clause(OSI)

source

public

core-gated

ungated

BSD-2-Clause license body confirmed. The repository is public and unarchived and builds the whole product, and the README describes no paid tier, enterprise edition or license-gated build beside it, so source is public and the core ungated. liger-kernel was the first product on the map to record BSD-2-Clause, and the shared software ladder's OSI tier lists license names literally, so the license mapped to no tier and the formula abstained. The tier now names it, on the owner's ruling, and the ladder reproduces this score rather than deferring it.

• https://github.com/linkedin/Liger-Kernel/blob/main/LICENSE recorded 2026-08-18

  LICENSE file is the verbatim BSD-2-Clause text

• https://api.github.com/repos/linkedin/Liger-Kernel recorded 2026-08-18

  Repo metadata - license spdx_id BSD-2-Clause, private false, archived false, default branch main - for linkedin/Liger-Kernel.

• https://github.com/linkedin/Liger-Kernel/blob/main/README.md recorded 2026-08-18

  README describes a collection of Triton kernels for LLM training that replace fused transformer operations and add memory-efficient post-training losses, applied by patching Hugging Face classes, with no paid tier, enterprise edition or license-key-gated build beside it.

584,477 PyPI downloads of `liger-kernel` in the trailing 30 days, which lands in the 100K-1M band of the software usage scale, level 3.

• https://pypistats.org/api/packages/liger-kernel/recent recorded 2026-08-18

  last_month downloads = 584,477 for liger-kernel

Banded on the category feature matrix as narrow kernel set or single-pass utility. Placed two bands below the tensorrt anchor, on a matrix that bands on how much of the model-to-hardware transformation pipeline a product performs, over how many inputs and targets.

• https://github.com/linkedin/Liger-Kernel/blob/main/README.md recorded 2026-08-18

  README still documents a collection of Triton kernels for LLM training that replace fused transformer operations and add memory-efficient post-training losses, applied by patching Hugging Face classes.