Merge to Learn: Efficiently Adding Skills to Language Models with Model Merging

Jacob Morrison, Noah Smith, Hannaneh Hajishirzi, Pang Wei Koh, Jesse Dodge, Pradeep Dasigi


Abstract
Adapting general-purpose language models to new skills is currently an expensive process that must be repeated as new instruction datasets targeting new skills are created, or can cause the models to forget older skills. In this work, we investigate the effectiveness of adding new skills to preexisting models by training on the new skills in isolation and later merging with the general model (e.g. using task vectors). In experiments focusing on scientific literature understanding, safety, and coding, we find that the parallel-train-then-merge procedure, which is significantly cheaper than retraining the models on updated data mixtures, is often comparably effective. Our experiments also show that parallel training is especially well-suited for enabling safety features in LMs relative to continued finetuning and retraining, as it dramatically improves model compliance with safe prompts while preserving its ability to refuse dangerous or harmful prompts.
Anthology ID:
2024.findings-emnlp.915
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:
15604–15621
Language:
URL:
https://aclanthology.org/2024.findings-emnlp.915
DOI:
Bibkey:
Cite (ACL):
Jacob Morrison, Noah Smith, Hannaneh Hajishirzi, Pang Wei Koh, Jesse Dodge, and Pradeep Dasigi. 2024. Merge to Learn: Efficiently Adding Skills to Language Models with Model Merging. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 15604–15621, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Merge to Learn: Efficiently Adding Skills to Language Models with Model Merging (Morrison et al., Findings 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.findings-emnlp.915.pdf