The Instruction Tax on Factual Recall: Base Models Can Beat Their Instruction-Tuned Counterparts
Paper recorded by Signals 4 on 2026-10-20 in cs.CL. Abstract reproduced from arXiv; link to the original below.
Published 2026-10-20 on arXiv · recorded by Signals 4 on 2026-10-21
Category: cs.CL · 自然语言处理 · first seen 2026-10-21
Abstract
We present our solution to the AKBC LM-KBC 2026 Shared Task, where we achieved a score of 0.6836 on the official leaderboard, more than double the baseline of 0.2964. This task requires generating all possible objects for subject-relation pairs using a language model with up to 32 billion parameters, with no retrieval or fine-tuning involved. We show that the prompting methodology is as important
Read on arXiv →
Cite this page: The Instruction Tax on Factual Recall: Base Models Can Beat Their Instruction-Tuned Counterparts: the #202 most recent of 660 cs.CL papers we have recorded (as of 2026-10-20). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/the-instruction-tax-on-factual-recall-base-models-can-beat-their-instruction-tun.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable:
papers.json
Get 4 AI signals a day by email — free.
Get 4 AI signals a day by email — free