PKU paper: rewriting influential training samples shifts model behavior more than reweighting

PKU · hf · 2026-09-10

• Paper from a PKU team: "From Reweighting to Rewriting: Unlocking the Intervention Effects of Influential Samples in Training Data Attribution."
• Key finding: influence-guided response rewriting of high-impact training examples (identified via training data attribution) produces stronger and more persistent behavioral shifts in language models than conventional reweighting.
• Takeaway: influential data offers broader intervention leverage — not just for explaining models, but as a more effective way to steer their behavior.

Original post →

More from Research

• Perturb the Physics, Not the Network: Magnetic Impurities Unlock Hamiltonian Learning — bravo_abad · 2026-09-10
• Vision-force fusion robot dressing handles moving arms: 85% arm coverage across 264 real trials — stepjamUK · 2026-09-10
• The J-lens Explained: Reading and Rewriting LLMs' Unspoken Concepts — CatAstro_Piyush · 2026-09-10
• Group Bench: ~100 group theory problems to benchmark your AI agents, with a dated progress map — Sauers_ · 2026-09-10
• Group Bench launches ~100 group theory problems for testing AI agents — Sauers_ · 2026-09-10
• ELLIS PhD Program Opens 2026 Applications With Cross-Border Co-Supervision — ArthurGretton · 2026-09-10

Research channel →