Active Fine-Tuning

Efficiently fine-tune large neural networks by intelligent active data selection.

 GitHub  Documentation  Paper

activeft retrieves data intelligently to maximize the information gain about specified prediction targets. This can be used to select data for efficient fine-tuning or to efficiently learn at test-time.

Publications

Efficiently Learning at Test-Time: Active Fine-Tuning of LLMs

Preprint

Transductive Active Learning: Theory and Applications

NeurIPS 2024