BobMcDear/attorch
A PyTorch-compatible neural network library written in Python using OpenAI's Triton for GPU kernel optimization.
Not currently ranked — collecting fresh signals.
star history
attorch reimplements PyTorch’s nn module using OpenAI’s Triton language for GPU acceleration. It provides optimized implementations of common neural network layers including attention mechanisms, convolutions, and normalization operations. The library supports both training and inference by implementing both forward and backward passes. It targets developers who need custom deep learning operations but lack expertise to write CUDA kernels directly.
Frequently asked
- What is BobMcDear/attorch?
- A PyTorch-compatible neural network library written in Python using OpenAI's Triton for GPU kernel optimization.
- Is attorch open source?
- Yes — BobMcDear/attorch is open source, released under the MIT license.
- What language is attorch written in?
- BobMcDear/attorch is primarily written in Python.
- How popular is attorch?
- BobMcDear/attorch has 605 stars on GitHub.
- Where can I find attorch?
- BobMcDear/attorch is on GitHub at https://github.com/BobMcDear/attorch.