OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Robustness Vit

Robustness Vit is a public research-code repository containing notebooks and analysis utilities for the paper Vision Transformers are Robust Learners. It examines Vision Transformer and convolutional-network performance under image corruptions, perturbations, distribution shifts and natural adversarial examples.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Robustness Vit — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you need research code and notebook workflows for investigating Vision Transformer robustness against image corruptions and distribution shifts.

Who it is for

  • Researchers studying image-classification robustness.
  • Practitioners who want to inspect notebook-based ImageNet robustness experiments.
  • Developers adapting the analysis to another model or dataset, with changes.

Documented uses

  • Explore reported model results across six ImageNet robustness datasets.
  • Inspect analyses of image masking and frequency-spectrum sensitivity.
  • Use the paper's notebooks as a starting point for a modified robustness evaluation.
Before you choose

Requirements & limits

The repository is public and MIT-licensed. The README identifies Python 3.8 as the development environment and says many ImageNet notebooks can be executed with Google Colab.

  • The repository is organized around research notebooks and analysis, not a hosted model-evaluation service.
  • Colab support varies by notebook; consult each notebook's instructions.
  • The authors describe a local research setup using GCP Vertex AI Workbench, four V100 GPUs, 60 GB RAM and 16 vCPUs; this is not presented as a universal minimum.
  • Adapting the analysis code for a different model or dataset requires modification.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Find the ImageNet robustness experiment notebooks

Locate the notebooks used to derive results for the six ImageNet robustness datasets.

Before you start

  • Access to the public GitHub repository

Steps

  1. Open the repository's imagenet_results directory.

  2. Choose a notebook for ImageNet-C, ImageNet-P, ImageNet-R, ImageNet-A, ImageNet-O or ImageNet-9.

  3. Check that notebook's execution instructions; the README says many, but not all, can be executed with Google Colab.

Expected outcome

A selected experiment notebook and its documented execution path.

Guide limitations

  • Colab execution is described as available for many notebooks, not all.
  • The README provides separate execution instructions when Colab is not applicable.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
sayakpaul/robustness-vit · Official homepage title · retained in its source language
How we review sourcesSuggest a correction