# EnzymeAD/Enzyme

High-performance automatic differentiation of LLVM and MLIR.

Repository: https://github.com/EnzymeAD/Enzyme
Canonical: https://ross.abutalabs.com/products/enzymead-enzyme
Homepage: https://enzyme.mit.edu
Language: LLVM
License: NOASSERTION
License Family: other
Topics: enzyme, llvm, ad, automatic-differentiation, c, cpp, tensorflow, pytorch, machine-learning, gradient, derivative, high-performance, rust, clang, deep-learning, scientific-computing, simulation, differentiable-programming, compiler, llvm-enzyme
Last push: 2026-08-26T12:44:46+00:00

## Health v2 (maintenance only)
Score: 95/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 100
- inputs: {"age_days": 2744, "days_push": 7, "days_rel": 8, "gap_med": 2.0, "n_releases_24m": 145}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1679, forks 184 (observed 2026-08-28T04:05:21.391992+00:00)

## What it is
Enzyme is a high-performance automatic differentiation plugin for LLVM and MLIR that computes derivatives and gradients of arbitrary existing code at the IR level. It works across many source languages (C, C++, Rust, Julia, Fortran, Swift, TensorFlow, PyTorch) by differentiating optimized code, often exceeding the performance of source-transformation AD tools.

## Use cases
- compute gradients of C/C++ code without rewriting it
- differentiate GPU kernels for machine learning
- generate fast reverse-mode gradients for scientific simulation
- add automatic differentiation to a custom compiler via LLVM/MLIR
- differentiate Rust or Julia programs with one tool
- speed up gradient computation compared to source-transformation AD

## When to avoid
- you only need Python-level autodiff for a pure PyTorch/JAX workflow where native autodiff suffices
- your code is not compiled to LLVM IR or uses dynamically unanalyzable constructs
- you want a pure-Python installable package with no LLVM toolchain dependency

## Facets
- artifact type: library
- maturity: active
- function: compiler, machine-learning, gpu-computing, developer-tools
- domain: machine-learning, compilers, deep-learning, performance
- platform: cpp, cross-platform, python, rust
- tags: automatic-differentiation, llvm-plugin, mlir, gradients, differentiable-programming, high-performance-computing, reverse-mode-ad, scientific-computing, llvm, gpu

## Member repositories
- EnzymeAD/Enzyme (main) score 95

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:21.391992+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:41:30.191830+00:00, confidence not recorded.
  - readme: https://github.com/EnzymeAD/Enzyme (fetched 2026-08-28T04:05:21.391992+00:00, sha 9693ce969e6b)
  - homepage: https://enzyme.mit.edu (fetched 2026-08-29T11:14:46.876052+00:00, sha f811f042c420)
  - site_page: https://enzyme.mit.edu/Installation (fetched 2026-08-29T11:14:46.887383+00:00, sha 5782a83370c5)
  - site_page: https://enzyme.mit.edu/getting_started/Faq (fetched 2026-08-29T11:14:46.885478+00:00, sha 10911fb9b130)
- Data as of 2026-08-30T08:39:29.467469+00:00.