Functionality learning through specification instructions

Pedro Henrique Luz De Araujo, Benjamin Roth


Abstract
Test suites assess natural language processing models’ performance on specific functionalities: cases of interest involving model robustness, fairness, or particular linguistic capabilities. This paper introduces specification instructions: text descriptions specifying fine-grained task-specific behaviors. For each functionality in a suite, we generate an instruction that describes it. We combine the specification instructions to create specification-augmented prompts, which we feed to language models pre-trained on natural instruction data.We conduct experiments to measure how optimizing for some functionalities may negatively impact functionalities that are not covered by the specification set. Our analyses across four tasks and models of diverse sizes and families show that smaller models struggle to follow specification instructions. However, larger models (> 3B params.) can benefit from specifications and—surprisingly—even generalize certain desirable behaviors across functionalities.
Anthology ID:
2024.findings-emnlp.642
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2024
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
10955–10990
Language:
URL:
https://preview.aclanthology.org/build-pipeline-with-new-library/2024.findings-emnlp.642/
DOI:
10.18653/v1/2024.findings-emnlp.642
Bibkey:
Cite (ACL):
Pedro Henrique Luz De Araujo and Benjamin Roth. 2024. Functionality learning through specification instructions. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 10955–10990, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Functionality learning through specification instructions (Luz De Araujo & Roth, Findings 2024)
Copy Citation:
PDF:
https://preview.aclanthology.org/build-pipeline-with-new-library/2024.findings-emnlp.642.pdf