Skip to content

@bethgelab Bethge Lab

Change the repository type filter

All

    Repositories list

    54 repositories

    • revenge-bench

      Public
      RevengeBench: Reverse Engineering Code-Space Policies from Behavioral Experiments
      Python
      •
      MIT License
      •0•1•0•0•Updated Jul 8, 2026Jul 8, 2026

    • qval

      Public
      Testbed to evaluate dense supervision methods.
      Python
      •
      MIT License
      •0•6•0•0•Updated Jul 1, 2026Jul 1, 2026

    • alignment-pattern-analysis

      Public
      Code for the ICLR 2026 paper "Only Brains Align with Brains: Cross-Region Alignment Patterns Expose Limits of Normative Models"
      Python
      •0•2•2•0•Updated Apr 23, 2026Apr 23, 2026

    • lit-benchmark

      Public
      Codebase for the LiT Benchmark from "Round-Trip Translation Reveals What Frontier Multilingual Benchmarks Miss"
      multilingualbenchmarkmultilingual-nlp
      multilingualbenchmarkmultilingual-nlp
      Python
      •
      MIT License
      •0•3•0•0•Updated Apr 22, 2026Apr 22, 2026

    • delta-belief-rl

      Public
      Official implementation of the ΔBelief-RL method.
      Python
      •
      MIT License
      •2•31•0•0•Updated Feb 28, 2026Feb 28, 2026

    • cabs

      Public
      [CVPR'26] Official codebase for "Concept-Aware Batch Sampling Improves Language-Image Pretraining"
      Python
      •
      MIT License
      •0•5•0•0•Updated Feb 21, 2026Feb 21, 2026

    • foolbox

      Public
      A Python toolbox to create adversarial examples that fool neural networks in PyTorch, TensorFlow, and JAX
      pythonmachine-learningtensorflow
      pythonmachine-learningtensorflowkeraspytorchadversarial-examplesadversarial-attacksjax
      Python
      •
      MIT License
      •439•3k•23•6•Updated Dec 3, 2025Dec 3, 2025

    • what-moves-the-eyes

      Public
      Project page for "What Moves the Eyes: Doubling Mechanistic Model Performance Using Deep Networks to Discover and Test Cognitive Hypotheses"
      Python
      •
      MIT License
      •0•1•0•0•Updated Nov 30, 2025Nov 30, 2025

    • supersanity

      Public
      A critical analysis of the Cambrian-S model and VSI-Super benchmarks
      Python
      •
      MIT License
      •0•16•2•0•Updated Nov 20, 2025Nov 20, 2025

    • sober-reasoning

      Public
      A Sober Look at Language Model Reasoning
      HTML
      •6•92•5•0•Updated Nov 18, 2025Nov 18, 2025

    • CiteME

      Public
      CiteME is a benchmark designed to test the abilities of language models in finding papers that are cited in scientific texts.
      lm-evaluationcitation-attributioncitation-dataset
      lm-evaluationcitation-attributioncitation-dataset
      Python
      •
      Other
      •5•48•0•0•Updated Nov 3, 2025Nov 3, 2025

    • onebench

      Public
      [ACL'25] The official code for "ONEBench to Test Them All: Sample-Level Benchmarking Over Open-Ended Capabilities"
      Python
      •
      MIT License
      •0•5•0•0•Updated Jul 29, 2025Jul 29, 2025

    • imagecorruptions

      Public
      Python package to corrupt arbitrary images.
      pythondata-augmentationpip3
      pythondata-augmentationpip3image-corruptions
      Python
      •
      Apache License 2.0
      •75•477•3•1•Updated May 6, 2025May 6, 2025

    • model-vs-human

      Public
      Benchmark your model on out-of-distribution datasets with carefully collected human comparison data (NeurIPS 2021 Oral)
      benchmarktensorflowpytorch
      benchmarktensorflowpytorchtoolboxrobustness
      Python
      •55•362•3•1•Updated Apr 17, 2025Apr 17, 2025

    • sort-and-search

      Public
      Code for the paper: "Efficient Lifelong Model Evaluation in an Era of Rapid Progress" [NeurIPS'24]
      Python
      •2•10•0•0•Updated Oct 17, 2024Oct 17, 2024

    • frequency_determines_performance

      Public
      Code for the paper: "No Zero-Shot Without Exponential Data: Pretraining Concept Frequency Determines Multimodal Model Performance" [NeurIPS'24]
      Jupyter Notebook
      •
      MIT License
      •5•94•1•0•Updated Apr 29, 2024Apr 29, 2024

    • bwki-weekly-tasks

      Public
      BWKI Task of the week
      Jupyter Notebook
      •3•1•0•2•Updated Mar 20, 2024Mar 20, 2024

    • DataTypeIdentification

      Public
      Code for the ICLR'24 paper: "Visual Data-Type Understanding does not emerge from Scaling Vision-Language Models"
      MIT License
      •0•13•0•0•Updated Jan 17, 2024Jan 17, 2024

    • robustness

      Public
      Robustness and adaptation of ImageNet scale models. Pre-Release, stay tuned for updates.
      machine-learningdomain-adaptation
      machine-learningdomain-adaptation
      Python
      •
      Apache License 2.0
      •6•138•1•1•Updated Jul 5, 2023Jul 5, 2023

    • game-of-noise

      Public
      Trained model weights, training and evaluation code from the paper "A simple way to make neural networks robust against diverse image corruptions"
      Python
      •
      MIT License
      •12•62•0•0•Updated May 24, 2023May 24, 2023

    • tools

      Public
      Shell
      •0•6•0•0•Updated Aug 3, 2022Aug 3, 2022

    • slurm-monitoring-public

      Public
      Monitor your high performance infrastructure configured over slurm using TIG stack
      pythoninfluxdbgrafana
      pythoninfluxdbgrafanaslurmtelegraf
      Python
      •
      MIT License
      •7•24•0•0•Updated Apr 26, 2022Apr 26, 2022

    • AnalysisBySynthesis

      Public
      Adversarially Robust Neural Network on MNIST.
      Python
      •
      Apache License 2.0
      •18•64•5•0•Updated Feb 4, 2022Feb 4, 2022

    • inference_results_v1.1

      Public
      Python
      •
      Other
      •24•0•0•0•Updated Jan 15, 2022Jan 15, 2022

    • slow_disentanglement

      Public
      Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Coding
      Jupyter Notebook
      •
      MIT License
      •12•73•2•0•Updated Nov 17, 2021Nov 17, 2021

    • notorious_difficulty_of_comparing_human_and_machine_perception

      Public
      Code for the three case studies: Closed Contour Detection, Synthetic Visual Reasoning Test, Recognition Gap
      Jupyter Notebook
      •
      MIT License
      •2•9•0•5•Updated Sep 14, 2021Sep 14, 2021

    • InDomainGeneralizationBenchmark

      Public
      Python
      •
      Apache License 2.0
      •6•35•0•0•Updated Aug 30, 2021Aug 30, 2021

    • stylize-datasets

      Public
      A script that applies the AdaIN style transfer method to arbitrary datasets
      Python
      •
      Other
      •38•166•1•1•Updated Jul 5, 2021Jul 5, 2021

    • testing_visualizations

      Public archive
      Code for the paper " Exemplary Natural Images Explain CNN Activations Better than Feature Visualizations"
      machine-learningdeep-learningcnn
      machine-learningdeep-learningcnnvisualizationsinterpretiability
      Python
      •2•10•0•0•Updated Jun 24, 2021Jun 24, 2021

    • openimages2coco

      Public
      Convert Open Images annotations into MS Coco format to make it a drop in replacement
      Jupyter Notebook
      •
      MIT License
      •20•116•3•0•Updated Aug 4, 2020Aug 4, 2020
    ProTip! When viewing an organization's repositories, you can use the props. filter to filter by custom property.