# pytorch/pytorch

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Repository: https://github.com/pytorch/pytorch
Canonical: https://ross.abutalabs.com/products/pytorch
Homepage: https://pytorch.org
Language: Python
License: NOASSERTION
License Family: other
Topics: neural-network, autograd, gpu, numpy, deep-learning, tensor, python, machine-learning
Last push: 2026-08-27T00:41:38+00:00

## Health v2 (maintenance only)
Score: 93/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 80, longevity 100
- inputs: {"age_days": 3672, "days_push": 7, "days_rel": 56, "gap_med": 50, "n_releases_24m": 14}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 102605, forks 28996 (observed 2026-08-28T04:12:23.714519+00:00)

## What it is
PyTorch is a Python library providing NumPy-like tensor computation with strong GPU acceleration and a tape-based automatic differentiation system for building deep neural networks. It supports dynamic computation graphs, distributed training, TorchScript for production deployment, and runs on CUDA, ROCm, Intel GPUs, and CPU.

## Use cases
- train deep neural networks in python
- gpu-accelerated tensor computation like numpy
- automatic differentiation for gradient descent
- build and train CNNs or transformers
- distributed multi-GPU model training
- prototype machine learning research models
- deploy models to production with torchscript

## When to choose
- you need a flexible, pythonic deep learning framework with dynamic graphs
- you want GPU-accelerated tensor math with autograd
- you need broad ecosystem support for research and production training
- you need distributed training across multiple GPUs or nodes

## When to avoid
- you only need lightweight NumPy-style arrays without GPU or autograd
- you need a fully static-graph framework for mobile or edge deployment with strict size limits
- you want a high-level API that hides training loops entirely

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning, llm-training, gpu-computing, math
- domain: machine-learning, deep-learning, artificial-intelligence, gpu-computing, data-science
- platform: python, windows, cross-platform
- tags: tensors, autograd, neural-networks, numpy, cuda, rocm, distributed-training, torchscript, linux, macos, gpu

## Member repositories
- pytorch/pytorch (main) score 93

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:23.714519+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:12:28.564137+00:00, confidence not recorded.
  - readme: https://github.com/pytorch/pytorch (fetched 2026-08-28T04:12:23.714519+00:00, sha 3dce22f96da6)
  - homepage: https://pytorch.org (fetched 2026-08-28T17:33:04.142685+00:00, sha 98fcafaad95b)
  - site_page: https://pytorch.org/docs/stable/index.html (fetched 2026-08-28T17:33:04.151537+00:00, sha a9a24195924a)
  - site_page: https://pytorch.org/docs (fetched 2026-08-28T17:33:04.155866+00:00, sha 57e35ca42d27)
  - site_page: https://pytorch.org/get-started/previous-versions (fetched 2026-08-28T17:33:04.153324+00:00, sha 0c42f6af76ca)
- Data as of 2026-08-30T08:39:29.467469+00:00.