Multiple paper open-source codes of the Microsoft Research Asia DKI group

A curated list of Composable AI methods: Building AI system by composing modules.

Distributional Generalization in NLP. A roadmap.

ReaSCAN is a synthetic navigation task that requires models to reason about surroundings over syntactically difficult languages. (NeurIPS '21)

Official pytorch implementation of CVPR2023 paper "Learning Conditional Attributes for Compositional Zero-Shot Learning"

The codebase for Inducing Causal Structure for Interpretable Neural Networks

ReCOGS: How Incidental Details of a Logical Form Overshadow an Evaluation of Semantic Interpretation

Diverse Demonstrations Improve In-context Compositional Generalization

[ACL'2023 Oral] "Learning to Substitute Span towards Improving Compositional Generalization"

Compositional Capabilities of Autoregressive Transformers: A Study on Synthetic, Interpretable Tasks

Code from the article: "Lost in Latent Space: Examining Failures of Disentangled Models at Combinatorial Generalisaton" (NeurIPS, 2022)

Official implementation of ICLR 2023 paper "A Minimalist Dataset for Systematic Generalization of Perception, Syntax, and Semantics"

This repository shares the most important sources used for the reasearch paper "More Diverse Training, Better Compositionality! Evidence from Multimodal Language Learning" by Caspar Volquardsen, Jae Hee Lee, Cornelius Weber, and Stefan Wermter.

Baby Abstract Reasoning Corpus (BabyARC) dataset engine, for generating grid-world-based abstract reasoning tasks on a large scale.

[ACL 2024] "Benchmarking and Improving Compositional Generalization of Multi-aspect Controllable Text Generation"

Living Survey for papers on Compositional Generalization in NLP

Causal Abstraction of Neural Models Trained to Solve ReaSCAN

Object-Centric Disentangled Mechanisms

Minimal model of tool discovery and tool innovation using active inference