Efficiently Learning at Test-Time: Active Fine-Tuning of LLMs
Keywords: data selection, retrieval, active learning, transductive active learning, local learning, test-time fine-tuning, transductive learning, language modeling, uncertainty quantification
TL;DR: We develop SIFT, an effective data selection method for fine-tuning LLMs. We show that test-time fine-tuning with SIFT can significantly and robustly improve language modeling ability.
Abstract: Recent efforts in fine-tuning language models often rely on automatic data selection, commonly using Nearest Neighbors retrieval from large datasets.
However, we theoretically show that this approach tends to select redundant data, limiting its effectiveness or even hurting performance.
To address this, we introduce SIFT, a data selection algorithm designed to reduce uncertainty about the model's response given a prompt, which unifies ideas from retrieval and active learning.
Whereas Nearest Neighbor retrieval typically fails in the presence of information duplication, SIFT accounts for information duplication and optimizes the overall information gain of the selected examples.
We focus our evaluations on fine-tuning at test-time for prompt-specific language modeling on the Pile dataset, and show that SIFT consistently outperforms Nearest Neighbor retrieval, with minimal computational overhead.
Moreover, we show that our uncertainty estimates can predict the performance gain of test-time fine-tuning, and use this to develop an adaptive algorithm that invests test-time compute proportional to realized performance gains.
We provide the `activeft` (Active Fine-Tuning) library which can be used as a drop-in replacement for Nearest Neighbor retrieval.
Supplementary Material: zip
Primary Area: foundation or frontier models, including LLMs
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 4576
Loading