https://github.com/apache/tvm
Open deep learning compiler stack for cpu, gpu and specialized accelerators
https://github.com/apache/tvm
compiler deep-learning gpu javascript machine-learning metal opencl performance rocm spirv tensor tvm vulkan
Last synced: 12 days ago
JSON representation
Open deep learning compiler stack for cpu, gpu and specialized accelerators
- Host: GitHub
- URL: https://github.com/apache/tvm
- Owner: apache
- License: apache-2.0
- Created: 2016-10-12T22:20:28.000Z (over 8 years ago)
- Default Branch: main
- Last Pushed: 2025-05-12T20:41:36.000Z (13 days ago)
- Last Synced: 2025-05-12T21:42:29.540Z (13 days ago)
- Topics: compiler, deep-learning, gpu, javascript, machine-learning, metal, opencl, performance, rocm, spirv, tensor, tvm, vulkan
- Language: Python
- Homepage: https://tvm.apache.org/
- Size: 107 MB
- Stars: 12,273
- Watchers: 375
- Forks: 3,587
- Open Issues: 140
-
Metadata Files:
- Readme: README.md
- License: LICENSE
- Codeowners: .github/CODEOWNERSHIP
Awesome Lists containing this project
- awesome-deep-model-compression - TVM (Apache) - glr-tutorials-frontend-from-tensorflow-py)] [](https://github.com/apache/tvm) (Tools / Hard-ware Integration)
- awesome-list - Apache TVM - Open deep learning compiler stack for cpu, gpu and specialized accelerators. (Deep Learning Framework / Deployment & Distribution)
- awesome-yolo-object-detection - TVM
- awesome-yolo-object-detection - TVM
- awesome-llmops - TVM - square) | (Performance / ML Compiler)
- awesome-cuda-and-hpc - TVM
README
Open Deep Learning Compiler Stack
==============================================
[Documentation](https://tvm.apache.org/docs) |
[Contributors](CONTRIBUTORS.md) |
[Community](https://tvm.apache.org/community) |
[Release Notes](NEWS.md)
Apache TVM is a compiler stack for deep learning systems. It is designed to close the gap between the
productivity-focused deep learning frameworks and the performance- and efficiency-focused hardware backends.
TVM works with deep learning frameworks to provide end-to-end compilation for different backends.
License
-------
TVM is licensed under the [Apache-2.0](LICENSE) license.
Getting Started
---------------
Check out the [TVM Documentation](https://tvm.apache.org/docs/) site for installation instructions, tutorials, examples, and more.
The [Getting Started with TVM](https://tvm.apache.org/docs/get_started/overview.html) tutorial is a great
place to start.
Contribute to TVM
-----------------
TVM adopts the Apache committer model. We aim to create an open-source project maintained and owned by the community.
Check out the [Contributor Guide](https://tvm.apache.org/docs/contribute/).
History and Acknowledgement
---------------------------
TVM started as a research project for deep learning compilation.
The first version of the project benefited a lot from the following projects:
- [Halide](https://github.com/halide/Halide): Part of TVM's TIR and arithmetic simplification module
originates from Halide. We also learned and adapted some parts of the lowering pipeline from Halide.
- [Loopy](https://github.com/inducer/loopy): use of integer set analysis and its loop transformation primitives.
- [Theano](https://github.com/Theano/Theano): the design inspiration of symbolic scan operator for recurrence.
Since then, the project has gone through several rounds of redesigns.
The current design is also drastically different from the initial design, following the
development trend of the ML compiler community.
The most recent version focuses on a cross-level design with TensorIR as the tensor-level representation
and Relax as the graph-level representation and Python-first transformations.
The project's current design goal is to make the ML compiler accessible by enabling most
transformations to be customizable in Python and bringing a cross-level representation that can jointly
optimize computational graphs, tensor programs, and libraries. The project is also a foundation
infra for building Python-first vertical compilers for domains, such as LLMs.