Can In-context Learners Learn a Reasoning Concept from Demonstrations?

Michal Štefánik, Marek Kadlčík


Abstract
Large language models show an emergent ability to learn a new task from a small number of input-output demonstrations. However, recent work shows that in-context learners largely rely on their pre-trained knowledge, such as the sentiment of the labels, instead of finding new associations in the input. However, the commonly-used few-shot evaluation settings using a random selection of in-context demonstrations can not disentangle models’ ability to learn a new skill from demonstrations, as most of the randomly-selected demonstrations do not present relations informative for prediction beyond exposing the new task distribution. To disentangle models’ in-context learning ability independent of models’ memory, we introduce a Conceptual few-shot learning method selecting the demonstrations sharing a possibly-informative concept with the predicted sample. We extract a set of such concepts from annotated explanations and measure how much can models benefit from presenting these concepts in few-shot demonstrations. We find that smaller models are more sensitive to the presented concepts. While some of the models are able to benefit from concept-presenting demonstrations for each assessed concept, we find that none of the assessed in-context learners can benefit from all presented reasoning concepts consistently, leaving the in-context concept learning an open challenge.
Anthology ID:
2023.nlrse-1.8
Volume:
Proceedings of the 1st Workshop on Natural Language Reasoning and Structured Explanations (NLRSE)
Month:
June
Year:
2023
Address:
Toronto, Canada
Editors:
Bhavana Dalvi Mishra, Greg Durrett, Peter Jansen, Danilo Neves Ribeiro, Jason Wei
Venue:
NLRSE
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
107–115
Language:
URL:
https://aclanthology.org/2023.nlrse-1.8
DOI:
10.18653/v1/2023.nlrse-1.8
Bibkey:
Cite (ACL):
Michal Štefánik and Marek Kadlčík. 2023. Can In-context Learners Learn a Reasoning Concept from Demonstrations?. In Proceedings of the 1st Workshop on Natural Language Reasoning and Structured Explanations (NLRSE), pages 107–115, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
Can In-context Learners Learn a Reasoning Concept from Demonstrations? (Štefánik & Kadlčík, NLRSE 2023)
Copy Citation:
PDF:
https://preview.aclanthology.org/naacl24-info/2023.nlrse-1.8.pdf
Video:
 https://preview.aclanthology.org/naacl24-info/2023.nlrse-1.8.mp4