← 研踪 PaperTrailarXiv:2611.03862 · v1
Dr. No: Feature Steering for Abstention in Vision-Language Models
首次提交 2026-11-03 03:34
版本更新 2026-11-03 03:34
雷达观测 2026-11-04 23:40
摘要
Large foundation models often answer even when the available visual or textual evidence is insufficient to support a reliable conclusion. This failure to abstain makes them hard to trust in high-stakes or information-seeking settings, where a confident but unsupported answer can be worse than no answer at all. It also raises a natural question: can the internal mechanisms behind this behaviour be identified and controlled at inference time? We present Dr. No, a feature-steering approach that mines sparse autoencoder (SAE) features associated with abstention and commitment, then modulates a small set of them during the prefill pass, without retraining the base model. Dr. No is evaluated on LLaVA-1.6 Mistral 7B across multimodal abstention benchmarks, including UNK-VQA and MM-UPD, and additionally benchmarked on language reasoning tasks using Qwen3.5-2B on AbstentionBench subsets. Against multiple uncertainty-based abstention baselines, feature steering achieves a more favourable balance between correct answering and appropriate abstention than post-hoc rejection methods. The gains are strongest on UNK-VQA and on MM-UPD cases involving image--question or question--answer incompatibility, and transfer most clearly to mathematical reasoning subsets of AbstentionBench. However, performance is more limited on absent-answer detection and retrieval-heavy multi-hop QA, such as MuSiQue. These results suggest that sparse internal features provide lightweight control handles for shaping uncertainty-aware behaviour in foundation models, while also showing that abstention control is source-dependent rather than governed by a single universal signal.
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-03 03:34 · 最新观测版本
Dr. No: Feature Steering for Abstention in Vision-Language Models
Large foundation models often answer even when the available visual or textual evidence is insufficient to support a reliable conclusion. This failure to abstain makes them hard to trust in high-stakes or information-seeking settings, where a confident but unsupported answer can be worse than no answer at all. It also raises a natural question: can the internal mechanisms behind this behaviour be identified and controlled at inference time? We present Dr. No, a feature-steering approach that mines sparse autoencoder (SAE) features associated with abstention and commitment, then modulates a small set of them during the prefill pass, without retraining the base model. Dr. No is evaluated on LLaVA-1.6 Mistral 7B across multimodal abstention benchmarks, including UNK-VQA and MM-UPD, and additionally benchmarked on language reasoning tasks using Qwen3.5-2B on AbstentionBench subsets. Against multiple uncertainty-based abstention baselines, feature steering achieves a more favourable balance between correct answering and appropriate abstention than post-hoc rejection methods. The gains are strongest on UNK-VQA and on MM-UPD cases involving image--question or question--answer incompatibility, and transfer most clearly to mathematical reasoning subsets of AbstentionBench. However, performance is more limited on absent-answer detection and retrieval-heavy multi-hop QA, such as MuSiQue. These results suggest that sparse internal features provide lightweight control handles for shaping uncertainty-aware behaviour in foundation models, while also showing that abstention control is source-dependent rather than governed by a single universal signal.
查看此版本解读 →查看此版本原文 ↗