mtsang / archipelago
Official Code Repo for the Paper: "How does This Interaction Affect Me? Interpretable Attribution for Feature Interactions", In NeurIPS 2020
☆36Updated last year
Related projects: ⓘ
- This is the official implementation for the paper "Learning to Scaffold: Optimizing Model Explanations for Teaching"☆19Updated 2 years ago
- Code for "Rissanen Data Analysis: Examining Dataset Characteristics via Description Length" by Ethan Perez, Douwe Kiela, and Kyungyhun Ch…☆35Updated 3 years ago
- Code for paper "Search Methods for Sufficient, Socially-Aligned Feature Importance Explanations with In-Distribution Counterfactuals"☆17Updated last year
- ☆83Updated last year
- Implementation of experiments in paper "Learning from Rules Generalizing Labeled Exemplars" to appear in ICLR2020 (https://openreview.net…☆49Updated last year
- Learning the Difference that Makes a Difference with Counterfactually-Augmented Data☆168Updated 3 years ago
- A Diagnostic Study of Explainability Techniques for Text Classification☆64Updated 3 years ago
- Pytorch implementation of DiffMask☆55Updated last year
- XAI-Bench is a library for benchmarking feature attribution explainability techniques☆55Updated last year
- ☆26Updated last year
- Interpretable Neural Predictions with Differentiable Binary Variables☆85Updated 3 years ago
- Interactive Weak Supervision: Learning Useful Heuristics for Data Labeling☆30Updated 3 years ago
- Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine Learning (AISTATS 2022 Oral)☆38Updated last year
- Code for paper "When Can Models Learn From Explanations? A Formal Framework for Understanding the Roles of Explanation Data"☆14Updated 3 years ago
- Code base for the EMNLP 2021 Findings paper: Cartography Active Learning☆14Updated last year
- ☆27Updated last year
- Code for EMNLP 2019 paper "Attention is not not Explanation"☆57Updated 3 years ago
- ☆24Updated 3 years ago
- This is a repository with the code for the EMNLP 2020 paper "Information-Theoretic Probing with Minimum Description Length"☆68Updated 3 weeks ago
- diagNNose is a Python library that facilitates a broad set of tools for analysing hidden activations of neural models.☆81Updated 10 months ago
- Code for Paper: Calibrated Language Model Fine-Tuning for In- and Out-of-Distribution Data☆35Updated 3 years ago
- Code for gradient rollback, which explains predictions of neural matrix factorization models, as for example used for knowledge base comp…☆20Updated 3 years ago
- Implementation of the "Poincare Glove: Hyperbolic word embeddings" paper☆84Updated 3 years ago
- Source code for "Towards Hierarchical Importance Attribution: Explaining Compositional Semantics for Neural Sequence Models", ICLR 2020.☆30Updated 4 years ago
- ☆87Updated 2 years ago
- Model zoo for different kinds of uncertainty quantification methods used in Natural Language Processing, implemented in PyTorch.☆45Updated last year
- ☆63Updated 2 years ago
- Algorithms for abstention, calibration and domain adaptation to label shift.☆36Updated 3 years ago
- Library implementing state-of-the-art Concept-based and Disentanglement Learning methods for Explainable AI☆52Updated 2 years ago
- Demo for method introduced in "Beyond Word Importance: Contextual Decomposition to Extract Interactions from LSTMs"☆56Updated 4 years ago