Aligning the Query Space: Greedy Information Projection for Language Model Data Selection
Greedy Information Projection is presented, a query-aligned method that requires only candidate embeddings and a score signal from LLM judgments, metadata, or intrinsic geometry, and which matches or surpasses full-data fine-tuning using small instruction and reasoning subsets.
Victor Ye Dong, Kuan-Yun Lee, Jia-Mei Shuai et al.
· 2 citations