PyTorch

Dynamic neural network framework from Meta — the dominant platform for ML research and increasingly for production.

Python free Open Source data since 2016

PyTorch’s dynamic computation graph (“define-by-run”) makes debugging neural networks intuitive — you use standard Python control flow and print statements. It became the dominant framework for ML research and, with TorchServe and TorchDeploy, is now widely used in production at Meta, Tesla, and OpenAI. Most modern LLMs are built and trained in PyTorch.

// tags

mlaideep-learningresearchgpupythonmetadynamic
Something wrong? Edit this entry on GitHub.
✏ Edit on GitHub