nlp:compositional_generalization

Table of Contents

Compositional Generalization

Compositional Generalization

Overviews

Lin et al 2023 - A Survey on Compositional Generalization in Applications Not an NLP paper, and not very comprehensive. WARNING: Missing a bunch of NLP work.

Papers

SCAN dataset: Lake & Baroni 2017 - Generalization without systematicity: On the compositional skills of sequence-to-sequence recurrent networks
Chang et al 2018 - Automatically Composing Representation Transformations as a Means for Generalization Has a good section on what compositional generalization is. Positive results for curriculum learning
Lake 2019 - Compositional generalization through metasequence-to-sequence learning
M-PCFGSET dataset: Hupkes et al 2019 - The Compositionality of Neural Networks: Integrating Symbolism and Connectionism
COGS dataset: Kim & Linzen 2020 - COGS: A Compositional Generalization Challenge Based on Semantic Interpretation
2019 - Measuring Compositional Generalization: A Comprehensive Method on Realistic Data Introduces MCD data split method
Chen et al 2020 - Compositional Generalization via Neural-Symbolic Stack Machines
Herzig & Berant 2020 - Span-based Semantic Parsing for Compositional Generalization
Shaw et al 2020 - Compositional Generalization and Natural Language Variation: Can a Semantic Parsing Approach Handle Both? Code and data Introduces TMCD split method
Improving Transformers for COG:
- Csordás et al 2021 - The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers
- Ontañón et al 2021 - Making Transformers Solve Compositional Tasks

Papers using Curriculum Learning

2020 - Compositional Generalization by Learning Analytical Expressions

Prompting and Language Models

Drozdov 2022 - Compositional Semantic Parsing with Large Language Models

COG in Machine Translation

Yin et al 2022 - Categorizing Semantic Representations for Neural Machine Translation

Non-NLP Papers

Klinger et al 2020 - A Study of Compositional Generalization in Neural Models Has some negative results for curriculum learning

Datasets

Semantic Parsing
- SCAN dataset paper. Also introduces a machine translation dataset for compositional generalization
- COGS dataset paper
- Spider dataset paper
CFQ dataset paper
PCFG dataset paper (A string edit operation composition benchmark)
M-PCFGSET dataset paper
Question Answering
- ComQA: Compositional Question-Answering dataset paper
Machine Translation
- Small-scale MT dataset from the SCAN dataset paper
- CoGnition dataset paper

People

Jacob Andreas Tw

Related Pages

Curriculum Learning (often useful for compositional generalization)
Effects of the Random Seed

nlp/compositional_generalization.txt · Last modified: 2023/07/07 18:21 by jmflanig