PyTorch
The dominant deep learning framework, built around dynamic computation graphs.
Overview
PyTorch is a deep learning framework whose graphs are built as the code runs, so ordinary Python control flow and debuggers work. It solves the problem that static graphs make a model hard to inspect and hard to change. It exists as the framework most research is published in, and that matters practically: a new technique usually appears as PyTorch code first, which is why production stacks tend to inherit it.
Technology Graph
How PyTorch connects to the rest of the stack.
Uses and depends on
- NVIDIA CUDA platform Hardware PyTorch uses NVIDIA CUDA High ✓ source
Applied to
- Deep Learning concept Algorithms PyTorch used for Deep Learning High ✓ source
Tools
Software that implements or integrates with PyTorch.