Infrastructure / Compilers & Model Optimization

FreeTensor

FreeTensor is a compiler and embedded Python DSL for writing and optimizing loop-based tensor programs, supporting dynamic tensor shapes, OpenMP and CUDA parallelization, explicit loop and memory-layout transformations, and reverse-mode automatic differentiation, compiling down to native code.

Verified 2026-09-02 via the GitHub API and the LICENSE body.

Openness

5 high confidence

5.0

license: Apache-2.0(OSI)
source: public
core-gated: ungated

Apache-2.0 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.

https://github.com/roastduck/FreeTensor/blob/master/LICENSE recorded 2026-09-02
LICENSE body carries the verbatim Apache License 2.0 text.
https://api.github.com/repos/roastduck/FreeTensor recorded 2026-09-02
Repo metadata for freetensor's canonical repository - private false, archived false, fork false.
https://github.com/roastduck/FreeTensor/blob/master/README.md recorded 2026-09-02
README describes FreeTensor as compiling loop-based tensor programs written in Python to native code, with dynamic shapes, parallelization, explicit optimization transformations and automatic differentiation, and names no paid tier, enterprise edition or license-gated build beside the published source.

Adoption

1 low confidence

1.0

152 GitHub stars, which lands in the <1K stars band of the stars scale, level 1. FreeTensor has no PyPI or other registry package and is built from source, so no download count exists and the stars scale applies.

https://api.github.com/repos/roastduck/FreeTensor recorded 2026-09-02
Repo metadata - stargazers_count = 152 - for roastduck/FreeTensor.

Capability

4 medium confidence

4.0

One band below the apache-tvm anchor, on the same reasoning as taco and Finch.jl: a full compilation toolchain - including automatic differentiation, which neither of those two provides - narrowed to one class of programs rather than apache-tvm's retargetable multi-framework, multi-vendor scope.

https://github.com/roastduck/FreeTensor/blob/master/README.md recorded 2026-09-02
README describes FreeTensor as compiling loop-based tensor programs written in Python to native code, with dynamic shapes, parallelization, explicit optimization transformations and automatic differentiation, and names no paid tier, enterprise edition or license-gated build beside the published source.

Verified 2026-09-02