apache/tvm-ffi
A minimal, open ABI and FFI specification for machine learning systems enabling cross-framework kernel interoperability.
Not currently ranked — collecting fresh signals.
star history
Apache TVM FFI defines a standardized foreign function interface for machine learning systems. It allows kernel libraries and domain-specific languages to be shipped once while supporting multiple ML frameworks and runtimes including PyTorch, JAX, and PaddlePaddle. The specification is framework-agnostic yet designed specifically to enable interoperability between compiler runtimes, kernel DSLs, and ML frameworks.
Frequently asked
- What is apache/tvm-ffi?
- A minimal, open ABI and FFI specification for machine learning systems enabling cross-framework kernel interoperability.
- Is tvm-ffi open source?
- Yes — apache/tvm-ffi is open source, released under the Apache-2.0 license.
- What language is tvm-ffi written in?
- apache/tvm-ffi is primarily written in C++.
- How popular is tvm-ffi?
- apache/tvm-ffi has 426 stars on GitHub.
- Where can I find tvm-ffi?
- apache/tvm-ffi is on GitHub at https://github.com/apache/tvm-ffi.