Measuring Pointwise 𝒱-Usable Information In-Context-ly

Sheng Lu, Shan Chen, Yingya Li, Danielle Bitterman, Guergana Savova, Iryna Gurevych


Abstract
In-context learning (ICL) is a new learning paradigm that has gained popularity along with the development of large language models. In this work, we adapt a recently proposed hardness metric, pointwise 𝒱-usable information (PVI), to an in-context version (in-context PVI). Compared to the original PVI, in-context PVI is more efficient in that it requires only a few exemplars and does not require fine-tuning. We conducted a comprehensive empirical analysis to evaluate the reliability of in-context PVI. Our findings indicate that in-context PVI estimates exhibit similar characteristics to the original PVI. Specific to the in-context setting, we show that in-context PVI estimates remain consistent across different exemplar selections and numbers of shots. The variance of in-context PVI estimates across different exemplar selections is insignificant, which suggests that in-context PVI estimates are stable. Furthermore, we demonstrate how in-context PVI can be employed to identify challenging instances. Our work highlights the potential of in-context PVI and provides new insights into the capabilities of ICL.
Anthology ID:
2023.findings-emnlp.1054
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2023
Month:
December
Year:
2023
Address:
Singapore
Editors:
Houda Bouamor, Juan Pino, Kalika Bali
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
15739–15756
Language:
URL:
https://aclanthology.org/2023.findings-emnlp.1054
DOI:
10.18653/v1/2023.findings-emnlp.1054
Bibkey:
Cite (ACL):
Sheng Lu, Shan Chen, Yingya Li, Danielle Bitterman, Guergana Savova, and Iryna Gurevych. 2023. Measuring Pointwise 𝒱-Usable Information In-Context-ly. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 15739–15756, Singapore. Association for Computational Linguistics.
Cite (Informal):
Measuring Pointwise 𝒱-Usable Information In-Context-ly (Lu et al., Findings 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.findings-emnlp.1054.pdf