ttnn

[Deprecated] ⭐️ TT-NN Compiler for PyTorch 2 ⭐️ Enables running PyTorch models on Tenstorrent hardware using eager or compile path

Meta's UMA and Orbital Materials' Orb-v3/OrbMol interatomic potential models, running on Tenstorrent hardware.

Publish and pull precompiled tt-metal kernel caches over Hugging Face Hub for Tenstorrent accelerators

Day-by-day performance tracking dashboard for Tenstorrent TT-Metal TTNN eltwise operations.

Web app to test and visualize 150+ Tenstorrent TTNN operations in real time, with PyTorch comparison

Cross-backend ML inference benchmarking tool — CPU/CUDA/TTNN. Designed for the Tenstorrent ML Benchmarking intern role.

Fused and block-float ML kernels for Tenstorrent Tensix cores in TT-Lang and TT-Metalium C++, measured on Tenstorrent's simulators, with CI on simulated chips, containers, a Kubernetes JobSet sweep and Ansible provisioning.

Interactive zero-dependency dashboard for the TTNN (Tenstorrent) op test matrix — every op across dtype × layout × memory, with per-config PCC/ULP accuracy and run-to-run diffs. Deploys to Cloudflare Workers.