#4444th of 56 in ML frameworks and libraries
tvm
Python-first machine learning compiler for optimizing and deploying models
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 13.8k
- License
- Apache-2.0
- Last commit
- Oct 2026
- Last release
- Sep 2026
- Language
- Python
Missed the latest GitHub check; these facts may be older.
Overview
Apache TVM is a machine learning compilation framework that uses TensorIR for tensor programs and Relax for computational graphs. Compiler transformations are customizable in Python, and the aim is to package models into minimal deployable modules. It also serves as a base for building domain-specific compilers, such as for LLMs.
Who it is for: ML compiler engineers and researchers optimizing model deployment
Strengths
- Cross-level design: TensorIR for tensor programs and Relax for graphs, optimized jointly
- Most compiler transformations can be customized in Python
- Apache-2.0 license under the Apache committer governance model
- Documentation site includes a getting-started tutorial and contributor guide
Weaknesses
- README lists no supported hardware targets or model formats
- README gives no install steps, version requirements or benchmarks
- Compiler workflow has a steeper learning curve than ready-made inference servers
- No ready-to-run model server or UI is described in the README
What it needs
- Docker
Also in ML frameworks and libraries
See all 56| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 95 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 94 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 92 out of 100 |
| #4 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 89 out of 100 |
| #5 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 89 out of 100 |