#7878th of 105 in ML frameworks and libraries
pyro
Probabilistic programming library for Bayesian modeling and inference on PyTorch
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 9.1k
- License
- Apache-2.0
- Last commit
- Oct 2026
- Last release
- Oct 2026
- Language
- Python
8,946 on 1 Apr 20269,068 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Pyro is a Python probabilistic programming library built on PyTorch. It lets you write generative models as Python programs and run inference on them, with high-level abstractions and hooks for customizing inference. Originally developed at Uber AI, it is now a Linux Foundation project maintained by community contributors, including a team at the Broad Institute.
Who it is for: Researchers and ML engineers doing Bayesian modeling in PyTorch
Strengths
- Built on PyTorch, so models can use tensors and neural networks directly
- Described as universal: can represent any computable probability distribution
- High-level inference abstractions, with easy access for customizing inference
- Linux Foundation project with published JMLR paper, docs and a community forum
Weaknesses
- Requires knowledge of probabilistic modeling and Bayesian inference to be useful
- Tied to PyTorch; no other tensor backend is mentioned in the README
- Example and tutorial models need the separate [extras] install
- README gives no hardware requirements or benchmark figures
What it needs
- GPU optional
- Docker
- Needs PyTorch
Also in ML frameworks and libraries
See all 105| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 94 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 94 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 93 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 92 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 91 out of 100 |