bytedance/byteir
A ByteDance model compilation solution using MLIR to optimize and lower deep learning models for various hardware backends.
Not currently ranked — collecting fresh signals.
star history
ByteIR provides compiler, runtime, and frontend components for end-to-end model compilation. It adopts upstream MLIR dialects and Google MHLO to lower state-of-the-art deep learning models and supports PyTorch, TensorFlow, and ONNX frontends. The project enables DL ASIC compilers to leverage generic graph, loop, and tensor-level optimizations while focusing on backend-specific code generation.
Frequently asked
- What is bytedance/byteir?
- A ByteDance model compilation solution using MLIR to optimize and lower deep learning models for various hardware backends.
- Is byteir open source?
- Yes — bytedance/byteir is open source, released under the Apache-2.0 license.
- What language is byteir written in?
- bytedance/byteir is primarily written in MLIR.
- How popular is byteir?
- bytedance/byteir has 471 stars on GitHub.
- Where can I find byteir?
- bytedance/byteir is on GitHub at https://github.com/bytedance/byteir.