AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Submodular Optimization for Minimal Augmentation in Robust Language Model Alignment

Ching-Chia Kao, Chia-Mu Yu, Chun-Shien Lu, Chu-song Chen

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Safety alignment after fine-tuning

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
Use submodular log-det selection instead of random selection (ablation)augmentation selection · Qwen2.5-7B harmful promptsASR: 17.6% -> 0.0%Improved
Use submodular log-det selection instead of Greedy Farthest Distance (ablation)augmentation selection · Qwen2.5-7B harmful promptsASR: 4.4% -> 0.0%Improved
Use submodular selection instead of Greedy Farthest on Mistral-7B (ablation)augmentation selection · Mistral-7B harmful promptsASR: 8.9 -> 0.0Improved
Use submodular selection instead of Greedy Farthest on Llama-3.1-8B (ablation)augmentation selection · Llama-3.1-8B harmful promptsASR: 13.3 -> 4.4Improved
Increase cyclic augmentation order from n=1 to n=2 (ablation)cyclic augmentation · Phi-3-mini inverse-alignment fine-tuningrobustness slope magnitude: -0.40 -> -0.21Improved
Increase cyclic augmentation order from n=1 to n=4 (ablation)cyclic augmentation · Phi-3-mini inverse-alignment fine-tuningrobustness slope magnitude: -0.40 -> -0.11Improved
Increase cyclic order n=1 to n=4 on Qwen-2.5-7B (ablation)cyclic augmentation · Qwen-2.5-7B inverse-alignment fine-tuningrobustness intercept R0: 0.73 -> 0.84Improved
Add third context to diminishing returns check, B and C vs empty context (ablation)augmentation A marginal gain · Phi-3-minimarginal gain of A: 0.25 -> 0.11Hurt
Use log-det surrogate with sentence embedding features (ablation)log-det surrogate · ranking quality of subsetsNDCG: 0.9701Improved
Use gradient embedding features instead of representation difference (ablation)log-det surrogate · ranking quality of subsetsSpearman: 0.2771 -> 0.0213Hurt
Sweep sigma from 0.01 to 1000 (ablation)log-det noise parameter · Qwen2.5-7B harmful promptsASR: 7.4% -> 4.4% -> 6.6%No clear effect
Use biased ground set Gbad instead of random ground set Grand1 with LOGDET (ablation)ground set choice · ground set ablationASR: 0.470 -> 0.565Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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