Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
auto_LiRPA: An Automatic Linear Relaxation based Perturbation Analysis Library for Neural Networks and General Computational Graphs
| Date | Stars |
|---|---|
| 2026-07-31 | 349 |
| 2026-08-05 | 350 |
| 2026-08-06 | 350 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# auto_LiRPA: Automatic Linear Relaxation based Perturbation Analysis for Neural Networks [](https://auto-lirpa.readthedocs.io/en/latest/?badge=latest) [](http://PaperCode.cc/AutoLiRPA-Demo) [](http://PaperCode.cc/AutoLiRPA-Video) [](https://opensource.org/licenses/BSD-3-Clause) <p align="center"> <a href="http://PaperCode.cc/AutoLiRPA-Video"><img src="http://www.huan-zhang.com/images/upload/lirpa/auto_lirpa_2.png" width="45%" height="45%" float="left"></a> <a href="http://PaperCode.cc/AutoLiRPA-Video"><img src="http://www.huan-zhang.com/images/upload/lirpa/auto_lirpa_1.png" width="45%" height="45%" float="right"></a> </p> ## What's New? - [α,β-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git) (using `auto_LiRPA` as its core library) is the winner of [VNN-COMP 2025](https://sites.google.com/view/vnn2025) and is **ranked top-1** in all [scored benchmarks](https://github.com/VNN-COMP/vnncomp2025_results/blob/main/SCORING-SMALL-TOL/latex/main.pdf). (08/2025) - Bounding of computation graphs containing Jacobian operators now supports more nonlinear operators (e.g., ```tanh```, ```sigmoid```), enabling verification of [continuous-time Lyapunov stability](https://github.com/Verified-Intelligence/Two-Stage_Neural_Controller_Training). (12/2025) - [α,β-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git) (using `auto_LiRPA` as its core library) is the winner of [VNN-COMP 2024](https://sites.google.com/view/vnn2024). Our tool is **ranked top-1** in all benchmarks (including 12 [regular track](https://github.com/ChristopherBrix/vnncomp2024_results/blob/main/SCORING/latex/results_regular_track.pdf) and 9 [extended track](https://github.com/ChristopherBrix/vnncomp2024_results/blob/main/SCORING/latex/results_extended_track.pdf) benchmarks). (08/2024) - The [INVPROP algorithm](https://arxiv.org/pdf/2302.01404.pdf) allows to compute overapproximationsw of preimages (the set of inputs of an NN generating a given output set) and tighten bounds using output constraints. (03/2024) - Branch-and-bound support for non-ReLU and general nonlinearities ([GenBaB](https://arxiv.org/pdf/2405.21063)) with optimizable bounds (α-CROWN) for new nonlinear functions (sin, cos, GeLU). We achieve significant improvements on verifying neural networks with non-ReLU nonlinearities such as Transformers, LSTM, and [ML4ACOPF](https://github.com/AI4OPT/ml4acopf_benchmark). (09/2023) - [α,β-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git) ([alpha-beta-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git)) (using `auto_LiRPA` as its core library) **won** [VNN-COMP 2023](https://sites.google.com/view/vnn2023). (08/2023) - Bound computation for higher-order computational graphs to support bounding Jacobian, Jacobian-vector products, and [local Lipschitz constants](https://arxiv.org/abs/2210.07394). (11/2022) - Our neural network verification tool [α,β-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git) ([alpha-beta-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git)) (using `auto_LiRPA` as its core library) **won** [VNN-COMP 2022](https://sites.google.com/view/vnn2022). Our library supports the large CIFAR100, TinyImageNet and ImageNet models in VNN-COMP 2022. (09/2022) - Implementation of **general cutting planes** ([GCP-CROWN](https://arxiv.org/pdf/2208.05740.pdf)), support of more activation functions and improved performance and scalability. (09/2022) - Our neural network verification tool [α,β-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git) ([alpha-beta-CROWN](https://github.com/Verified-Intelligence/alpha-beta-CROWN.git)) **won** [VNN-COMP 2
Excerpt of 19,001 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:d1013f5041b3d446, llm:Repository description: "auto_LiRPA: An Automatic Linear Relaxation based Perturbation Analysis Library for Neural Networks and General Computational Graphs" and topics: adversarial-defense, adversarial-examples, adversarial-machine-learning, certified-defense-models, crown-bound, ibp-bounds, robustness, robustness-verification.
matched fp:d1013f5041b3d446, llm:Repository description: "auto_LiRPA: An Automatic Linear Relaxation based Perturbation Analysis Library for Neural Networks and General Computational Graphs" and topics: adversarial-defense, adversarial-examples, adversarial-machine-learning, certified-defense-models, crown-bound, ibp-bounds, robustness, robustness-verification.