Nothing Special   »   [go: up one dir, main page]

Hit the Nail on the Head: Parameter-Efficient Multi-task Tuning via Human Language Intervention

Wenxuan Lu, Songhao Jiang, Wang Yijing, Tianning Zang


Abstract
Parameter-Efficient Fine-Tuning (PEFT) on small Pre-trained Language Models (PLMs) has emerged as a promising approach to enhance their multi-tasking capabilities. Prevalent methods simultaneously train additional modules (i.e., one task-shared module and multiple task-specific modules) for adapting PLMs to downstream tasks. However, their adaptability to new tasks is constrained, as the task-specific modules independently adapt to each task, overlooking the potential for knowledge transfer across tasks. In this paper, we propose a novel multi-task learning framework, Inspirational Pointer (IP), that enables the transfer of prior knowledge across tasks through human language intervention. Specifically, we attach task descriptions to the input samples, which are then mapped to corresponding task embeddings. Based on those embeddings, we adapt PLMs for downstream tasks. Similar tasks share akin descriptions, allowing new task samples close to similar trained tasks in the task embedding space, hitting the memory about trained tasks of the model. Our experiments on the T5 model demonstrate performance improvements of our method in multi-task learning and few-shot transfer learning. Further, we implemented the IP in decoder-only models including GPT2 and large language models (LLMs), and the results show that IP enhances the capabilities of decoder-only models.
Anthology ID:
2024.findings-emnlp.518
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:
8873–8885
Language:
URL:
https://aclanthology.org/2024.findings-emnlp.518
DOI:
10.18653/v1/2024.findings-emnlp.518
Bibkey:
Cite (ACL):
Wenxuan Lu, Songhao Jiang, Wang Yijing, and Tianning Zang. 2024. Hit the Nail on the Head: Parameter-Efficient Multi-task Tuning via Human Language Intervention. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 8873–8885, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Hit the Nail on the Head: Parameter-Efficient Multi-task Tuning via Human Language Intervention (Lu et al., Findings 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.findings-emnlp.518.pdf
Software:
 2024.findings-emnlp.518.software.zip
Data:
 2024.findings-emnlp.518.data.zip