AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Beyond URLs: Metadata Diversity and Position for Efficient LLM Pretraining

Dongyang Fan, Diba Hashemi, Sai Praneeth Karimireddy, Martin Jaggi

ICLR 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
prepend URL metadata1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 47.7Improved
prepend fine-grained quality score1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 47.3Improved
prepend fine-grained domain information1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 47.3Improved
prepend coarse-grained quality score1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 46.6No clear effect
prepend coarse-grained domain information1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 46.7No clear effect
prepend five learnable meta tokens1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 47.1Improved
append fine-grained domain information1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 47.3Improved
append coarse-grained quality score1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 47.1Improved
append URL1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 46.8No clear effect
append fine-grained quality score1.5B Llama · 9-task downstream averageaverage accuracy (%): 46.7 -> 46.8No clear effect
prepend full URL instead of only URL prefix (ablation)URL prepending · 9-task downstream averageaverage accuracy (%): 46.6 -> 47.7Improved
prepend full URL instead of only URL domain (ablation)URL prepending · 9-task downstream averageaverage accuracy (%): 47.2 -> 47.7Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy