@inproceedings{hosseini-etal-2022-compositional,
title = "On the Compositional Generalization Gap of In-Context Learning",
author = "Hosseini, Arian and
Vani, Ankit and
Bahdanau, Dzmitry and
Sordoni, Alessandro and
Courville, Aaron",
editor = "Bastings, Jasmijn and
Belinkov, Yonatan and
Elazar, Yanai and
Hupkes, Dieuwke and
Saphra, Naomi and
Wiegreffe, Sarah",
booktitle = "Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP",
month = dec,
year = "2022",
address = "Abu Dhabi, United Arab Emirates (Hybrid)",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.blackboxnlp-1.22",
doi = "10.18653/v1/2022.blackboxnlp-1.22",
pages = "272--280",
abstract = "Pretrained large generative language models have shown great performance on many tasks, but exhibit low compositional generalization abilities. Scaling such models has been shown to improve their performance on various NLP tasks even just by conditioning them on a few examples to solve the task without any fine-tuning (also known as in-context learning). In this work, we look at the gap between the in-distribution (ID) and out-of-distribution (OOD) performance of such models in semantic parsing tasks with in-context learning. In the ID settings, the demonstrations are from the same split ($\textit{test}$ or $\textit{train}$) that the model is being evaluated on, and in the OOD settings, they are from the other split. We look at how the relative generalization gap of in-context learning evolves as models are scaled up. We evaluate four model families, OPT, BLOOM, CodeGen and Codex on three semantic parsing datasets, CFQ, SCAN and GeoQuery with different number of exemplars, and observe a trend of decreasing relative generalization gap as models are scaled up.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hosseini-etal-2022-compositional">
<titleInfo>
<title>On the Compositional Generalization Gap of In-Context Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Arian</namePart>
<namePart type="family">Hosseini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ankit</namePart>
<namePart type="family">Vani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dzmitry</namePart>
<namePart type="family">Bahdanau</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alessandro</namePart>
<namePart type="family">Sordoni</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aaron</namePart>
<namePart type="family">Courville</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jasmijn</namePart>
<namePart type="family">Bastings</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yonatan</namePart>
<namePart type="family">Belinkov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanai</namePart>
<namePart type="family">Elazar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dieuwke</namePart>
<namePart type="family">Hupkes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Naomi</namePart>
<namePart type="family">Saphra</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sarah</namePart>
<namePart type="family">Wiegreffe</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, United Arab Emirates (Hybrid)</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Pretrained large generative language models have shown great performance on many tasks, but exhibit low compositional generalization abilities. Scaling such models has been shown to improve their performance on various NLP tasks even just by conditioning them on a few examples to solve the task without any fine-tuning (also known as in-context learning). In this work, we look at the gap between the in-distribution (ID) and out-of-distribution (OOD) performance of such models in semantic parsing tasks with in-context learning. In the ID settings, the demonstrations are from the same split (test or train) that the model is being evaluated on, and in the OOD settings, they are from the other split. We look at how the relative generalization gap of in-context learning evolves as models are scaled up. We evaluate four model families, OPT, BLOOM, CodeGen and Codex on three semantic parsing datasets, CFQ, SCAN and GeoQuery with different number of exemplars, and observe a trend of decreasing relative generalization gap as models are scaled up.</abstract>
<identifier type="citekey">hosseini-etal-2022-compositional</identifier>
<identifier type="doi">10.18653/v1/2022.blackboxnlp-1.22</identifier>
<location>
<url>https://aclanthology.org/2022.blackboxnlp-1.22</url>
</location>
<part>
<date>2022-12</date>
<extent unit="page">
<start>272</start>
<end>280</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T On the Compositional Generalization Gap of In-Context Learning
%A Hosseini, Arian
%A Vani, Ankit
%A Bahdanau, Dzmitry
%A Sordoni, Alessandro
%A Courville, Aaron
%Y Bastings, Jasmijn
%Y Belinkov, Yonatan
%Y Elazar, Yanai
%Y Hupkes, Dieuwke
%Y Saphra, Naomi
%Y Wiegreffe, Sarah
%S Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP
%D 2022
%8 December
%I Association for Computational Linguistics
%C Abu Dhabi, United Arab Emirates (Hybrid)
%F hosseini-etal-2022-compositional
%X Pretrained large generative language models have shown great performance on many tasks, but exhibit low compositional generalization abilities. Scaling such models has been shown to improve their performance on various NLP tasks even just by conditioning them on a few examples to solve the task without any fine-tuning (also known as in-context learning). In this work, we look at the gap between the in-distribution (ID) and out-of-distribution (OOD) performance of such models in semantic parsing tasks with in-context learning. In the ID settings, the demonstrations are from the same split (test or train) that the model is being evaluated on, and in the OOD settings, they are from the other split. We look at how the relative generalization gap of in-context learning evolves as models are scaled up. We evaluate four model families, OPT, BLOOM, CodeGen and Codex on three semantic parsing datasets, CFQ, SCAN and GeoQuery with different number of exemplars, and observe a trend of decreasing relative generalization gap as models are scaled up.
%R 10.18653/v1/2022.blackboxnlp-1.22
%U https://aclanthology.org/2022.blackboxnlp-1.22
%U https://doi.org/10.18653/v1/2022.blackboxnlp-1.22
%P 272-280
Markdown (Informal)
[On the Compositional Generalization Gap of In-Context Learning](https://aclanthology.org/2022.blackboxnlp-1.22) (Hosseini et al., BlackboxNLP 2022)
ACL
- Arian Hosseini, Ankit Vani, Dzmitry Bahdanau, Alessandro Sordoni, and Aaron Courville. 2022. On the Compositional Generalization Gap of In-Context Learning. In Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP, pages 272–280, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics.