TT-Buda is a software stack that can compile AI/ML models from several different frameworks, and execute them on Tenstorrent hardware.
The TT-Buda software stack can compile AI/ML models from several different frameworks such as PyTorch and Tensorflow, and execute them in many different ways on Tenstorrent hardware.
Note on terminology:
TT-Buda is the official Tenstorrent AI/ML compiler stack and PyBuda is the Python interface for TT-Buda. PyBuda allows users to access and utilize TT-Buda's features directly from Python. This includes directly importing model architectures and weights from PyTorch, TensorFlow, ONNX, and TFLite.
Model demos are now part of a separate repo:
https://github.com/tenstorrent/tt-buda-demos
See: Docs
See: TT-Buda Installation
Set LD_LIBRARY_PATH
to the location of third_party/budabackend/build/lib
- preferrably the absolute path to allow scripts to find them from anywhere.
See README.silicon.md for details on how to run on silicon.