OSS Tanbou
View on GitHub

PyTorch — tensor computation and automatic differentiation from Python

Discovery score 351.6OSS health 96
About these scores

Discovery score is an unbounded weighted, log-compressed index of stars, watchers, forks, and contributors. Growth momentum is its change over the observed period; OSS health is a 0–100 score from available repository recency, Community Health, and release data.

Stars
102,960
Primary language
Python
License
Not determined
Repository last updated
Sep 13, 2026

Overview

PyTorch is a Python package that extends NumPy-like tensor work to accelerators and uses tape-based automatic differentiation to build neural networks. It fits into the wider Python scientific stack while leaving developers direct control over model and training code.

Based on official documentation; not hands-on tested · Content checked:

What it controls

Its core is tensor computation plus autograd. Operations recorded on tensors provide gradients, which can be combined with neural-network modules, optimization, data loading, and distributed components. The README presents Python as the primary interface while retaining paths for C++ and CUDA extensions when needed.

Sources: [1]

Where it fits

It is a strong candidate for research code, model experiments that require close inspection, and teams that want to control custom training loops in Python. Projects already using NumPy, SciPy, or Cython can keep data preparation and model work in a familiar ecosystem.

Sources: [1]

Boundaries to review

Installation depends on the target: CPU, NVIDIA CUDA, AMD ROCm, and Intel GPU paths have different requirements. Merely having a GPU does not accelerate every workload; the compatible build, drivers, device placement, and available memory still matter. Source builds have substantial prerequisites, so start with the official environment-specific binary guidance.

Sources: [1]

Official sources

  1. [1]PyTorch README(2026-09-12)
Supplemental curator note

Selected for connecting familiar Python and NumPy concepts to GPU deep learning through extensive official documentation and a broad ecosystem.

Growth trends

Last 30 days

102,960 Stars

Trend data is still being collected.

Built with

  • Python

Categories and tags

Categories

Tags

GitHub Topics

  • neural-network
  • autograd
  • gpu
  • numpy
  • deep-learning
  • tensor
  • python
  • machine-learning
GitHub dataView detailed GitHub data
View on GitHub
Stars
102,960
Forks
29,211
Watchers
1,802
Open issues
14,080
Contributors
326
Primary language
Python
License
Not determined
Repository last updated
Sep 13, 2026
Report incorrect information

Tell us if any listing information is incorrect or outdated.