tt-metal
This neural network library integrates Python and C++ for use with hardware such as the e150 and n150. It supports a range of models, including LLMs, CNNs, and NLPs, providing enhancements via Tensor and Data Parallelism. See API references and programming guides specific to Tenstorrent's hardware, and review model demos like Falcon7B and Mistral for increased throughput. TT-Metalium is available for kernel development, aided by technical reports and examples. These resources update regularly, catering to ongoing advancements in deep learning.
