Abstract
Given a Large Language Model (LLM) generation, how can we identify which training data led to this generation? In this paper, we proposed RapidIn, a scalable framework adapting to LLMs for estimating the influence of each training data. The proposed framework consists of two stages: caching and retrieval. First, we compress the gradient vectors by over 200,000x, allowing them to be cached on disk or in GPU/CPU memory. Then, given a generation, RapidIn efficiently traverses the cached gradients to estimate the influence within minutes, achieving over a 6,326x speedup. Moreover, RapidIn supports multi-GPU parallelization to substantially accelerate caching and retrieval. Our empirical result confirms the efficiency and effectiveness of RapidIn.- Anthology ID:
- 2024.acl-long.48
- Volume:
- Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- August
- Year:
- 2024
- Address:
- Bangkok, Thailand
- Editors:
- Lun-Wei Ku, Andre Martins, Vivek Srikumar
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 841–860
- Language:
- URL:
- https://preview.aclanthology.org/icon-24-ingestion/2024.acl-long.48/
- DOI:
- 10.18653/v1/2024.acl-long.48
- Cite (ACL):
- Huawei Lin, Jikai Long, Zhaozhuo Xu, and Weijie Zhao. 2024. Token-wise Influential Training Data Retrieval for Large Language Models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 841–860, Bangkok, Thailand. Association for Computational Linguistics.
- Cite (Informal):
- Token-wise Influential Training Data Retrieval for Large Language Models (Lin et al., ACL 2024)
- PDF:
- https://preview.aclanthology.org/icon-24-ingestion/2024.acl-long.48.pdf