# BorealisAI/advertorch

A Toolbox for Adversarial Robustness Research

Repository: https://github.com/BorealisAI/advertorch
Canonical: https://ross.abutalabs.com/products/advertorch
Language: Jupyter Notebook
License: LGPL-3.0
License Family: copyleft
Topics: pytorch, adversarial-examples, adversarial-example, adversarial-perturbations, adversarial-attacks, adversarial-machine-learning, adversarial-learning, robustness, toolbox, security, machine-learning, benchmarking
Last push: 2023-09-14T02:51:02+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2834, "days_push": 1084, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1364, forks 196 (observed 2026-08-28T04:04:30.958147+00:00)

## What it is
AdverTorch is a Python toolbox for adversarial robustness research built on PyTorch. It provides modules for generating adversarial perturbations, defending against adversarial examples, and scripts for adversarial training.

## Use cases
- generate adversarial examples to attack a pytorch model
- run PGD attacks on image classifiers
- adversarially train a robust model on MNIST
- evaluate robustness of machine learning models against perturbations
- implement defenses against adversarial examples
- benchmark adversarial attacks against foolbox or cleverhans implementations

## When to choose
- you research adversarial machine learning in PyTorch
- you need ready-made attacks like LinfPGDAttack with targeted/untargeted modes
- you want tutorial notebooks on attack/defense and adversarial training

## When to avoid
- you need TensorFlow or other framework support
- you need actively maintained tooling with recent PyTorch versions
- you need production security tooling rather than research code

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, security, benchmarking, testing
- domain: machine-learning, security, deep-learning, artificial-intelligence
- platform: python
- tags: adversarial-attacks, adversarial-robustness, pytorch, adversarial-training, adversarial-examples, research-toolbox, gpu

## Member repositories
- BorealisAI/advertorch (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:30.958147+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 2026-08-30T04:41:17.604802+00:00, confidence not recorded.
  - readme: https://github.com/BorealisAI/advertorch (fetched 2026-08-28T04:04:30.958147+00:00, sha afaf33cf4ca5)
  - registry_pypi: https://pypi.org/pypi/advertorch/json (fetched 2026-08-29T11:58:34.453074+00:00, sha f55d97b81f59)
- Data as of 2026-08-30T08:39:29.467469+00:00.