Vibe Coding Discovertt-metal
View on GitHub:metal: TT-NN operator library, and TT-Metalium low level kernel programming model.
TT-Metal provides TT-NN, a Python and C++ neural-network operator library, plus TT-Metalium for low-level kernel development on Tenstorrent accelerators. It includes optimized model implementations and multi-device inference support.
Use Cases
Run LLM inference on Tenstorrent hardwareDevelop low-level accelerator kernelsOptimize neural network operatorsScale inference across multiple devicesRun speech recognition models
Built With
- Language
- Python
- Frameworks
- TT-NN · TT-Metalium · vLLM
Tags
AI accelerator · neural network operators · GPU kernels · LLM inference · distributed inference · model optimization · Tenstorrent