Clarify or Answer: Reinforcement Learning for Agentic VQA with Context Under-specification

• URL: http://arxiv.org/abs/2601.16400v1
• Date: Fri, 23 Jan 2026 02:12:33 GMT
• Title: Clarify or Answer: Reinforcement Learning for Agentic VQA with Context Under-specification
• Abstract summary: We propose CoA, an ask-or-answer agent that separately models the decision to ask or answer, and what to ask if needed.<n>CoA first determines whether clarification is necessary; if so, it asks a single focused question and then incorporates the response to produce the final answer.<n>CoA achieves consistent improvements at both the module and system levels, improving end-to-end VQA accuracy by an average of +15.3 points (83%) over prompting-based baselines.
• Score: 10.412378789389544
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Real-world visual question answering (VQA) is often context-dependent: an image-question pair may be under-specified, such that the correct answer depends on external information that is not observable in the image. In such cases, directly answering can lead to confident but incorrect predictions. We propose CoA(Clarify-or-Answer), an ask-or-answer agent that separately models the decision to ask or answer, and what to ask if needed. CoA first determines whether clarification is necessary; if so, it asks a single focused question and then incorporates the response to produce the final answer. We introduce CONTEXTCLARIFY with a set of ambiguous VQA questions and the contrast set that is non-ambiguous. We further introduce GRPO-CR (Clarification Reasoning), a reinforcement learning approach that optimizes clarification question generation with multiple reward signals encouraging well-formed, focused, non-trivial questions that resolve ambiguity. Across three VLLMs and three datasets, CoA achieves consistent improvements at both the module and system levels, improving end-to-end VQA accuracy by an average of +15.3 points (83%) over prompting-based baselines

Related papers

• Clarification Is Not Enough: Post-Clarification Answering Remains the Bottleneck in Multi-Turn QA [14.475086992410276]
  We study the problem of preference elicitation in multi-turn question answering by decomposing the problem into two components.<n>We show, using the PACIFIC benchmark, that supervised fine-tuning rapidly improves the clarification policy, however, final answer accuracy remains substantially lower even when the model takes the correct action.
  arXiv  Detail & Related papers  (2026-05-24T18:06:09Z)
• Inferential Question Answering [67.54465021408724]
  We introduce Inferential QA -- a new task that challenges models to infer answers from answer-supporting passages which provide only clues.<n>To study this problem, we construct QUIT (QUestions requiring Inference from Texts) dataset, comprising 7,401 questions and 2.4M passages.<n>We show that methods effective on traditional QA tasks struggle in inferential QA: retrievers underperform, rerankers offer limited gains, and fine-tuning provides inconsistent improvements.
  arXiv  Detail & Related papers  (2026-02-01T14:02:43Z)
• Improving Zero-shot Visual Question Answering via Large Language Models with Reasoning Question Prompts [22.669502403623166]
  We present Reasoning Question Prompts for VQA tasks, which can further activate the potential of Large Language Models. We generate self-contained questions as reasoning question prompts via an unsupervised question edition module. Each reasoning question prompt clearly indicates the intent of the original question. Then, the candidate answers associated with their confidence scores acting as answer integritys are fed into LLMs.
  arXiv  Detail & Related papers  (2023-11-15T15:40:46Z)
• Davidsonian Scene Graph: Improving Reliability in Fine-grained Evaluation for Text-to-Image Generation [64.64849950642619]
  We develop an evaluation framework inspired by formal semantics for evaluating text-to-image models. We show that Davidsonian Scene Graph (DSG) produces atomic and unique questions organized in dependency graphs. We also present DSG-1k, an open-sourced evaluation benchmark that includes 1,060 prompts.
  arXiv  Detail & Related papers  (2023-10-27T16:20:10Z)
• UNK-VQA: A Dataset and a Probe into the Abstention Ability of Multi-modal Large Models [55.22048505787125]
  This paper contributes a comprehensive dataset, called UNK-VQA. We first augment the existing data via deliberate perturbations on either the image or question. We then extensively evaluate the zero- and few-shot performance of several emerging multi-modal large models.
  arXiv  Detail & Related papers  (2023-10-17T02:38:09Z)
• Open-Set Knowledge-Based Visual Question Answering with Inference Paths [79.55742631375063]
  The purpose of Knowledge-Based Visual Question Answering (KB-VQA) is to provide a correct answer to the question with the aid of external knowledge bases. We propose a new retriever-ranker paradigm of KB-VQA, Graph pATH rankER (GATHER for brevity) Specifically, it contains graph constructing, pruning, and path-level ranking, which not only retrieves accurate answers but also provides inference paths that explain the reasoning process.
  arXiv  Detail & Related papers  (2023-10-12T09:12:50Z)
• Selectively Answering Ambiguous Questions [38.83930394700588]
  We find that the most reliable approach to decide when to abstain involves quantifying repetition within sampled model outputs. Our results suggest that sampling-based confidence scores help calibrate answers to relatively unambiguous questions.
  arXiv  Detail & Related papers  (2023-05-24T01:25:38Z)
• Invariant Grounding for Video Question Answering [72.87173324555846]
  Video Question Answering (VideoQA) is the task of answering questions about a video. In leading VideoQA models, the typical learning objective, empirical risk minimization (ERM), latches on superficial correlations between video-question pairs and answers. We propose a new learning framework, Invariant Grounding for VideoQA (IGV), to ground the question-critical scene.
  arXiv  Detail & Related papers  (2022-06-06T04:37:52Z)
• Co-VQA : Answering by Interactive Sub Question Sequence [18.476819557695087]
  This paper proposes a conversation-based VQA framework, which consists of three components: Questioner, Oracle, and Answerer. To perform supervised learning for each model, we introduce a well-designed method to build a SQS for each question on VQA 2.0 and VQA-CP v2 datasets.
  arXiv  Detail & Related papers  (2022-04-02T15:09:16Z)
• Double Retrieval and Ranking for Accurate Question Answering [120.69820139008138]
  We show that an answer verification step introduced in Transformer-based answer selection models can significantly improve the state of the art in Question Answering. The results on three well-known datasets for AS2 show consistent and significant improvement of the state of the art.
  arXiv  Detail & Related papers  (2022-01-16T06:20:07Z)
• Estimating semantic structure for the VQA answer space [6.49970685896541]
  We show that our approach is completely model-agnostic since it allows consistent improvements with three different VQA models. We report SOTA-level performance on the challenging VQAv2-CP dataset.
  arXiv  Detail & Related papers  (2020-06-10T08:32:56Z)
• Rephrasing visual questions by specifying the entropy of the answer distribution [0.0]
  We propose a novel task, rephrasing the questions by controlling the ambiguity of the questions. The ambiguity of a visual question is defined by the use of the entropy of the answer distribution predicted by a VQA model. We demonstrate the advantage of our approach that can control the ambiguity of the rephrased questions, and an interesting observation that it is harder to increase than to reduce ambiguity.
  arXiv  Detail & Related papers  (2020-04-10T09:32:37Z)
• SQuINTing at VQA Models: Introspecting VQA Models with Sub-Questions [66.86887670416193]
  We show that state-of-the-art VQA models have comparable performance in answering perception and reasoning questions, but suffer from consistency problems. To address this shortcoming, we propose an approach called Sub-Question-aware Network Tuning (SQuINT) We show that SQuINT improves model consistency by 5%, also marginally improving performance on the Reasoning questions in VQA, while also displaying better attention maps.
  arXiv  Detail & Related papers  (2020-01-20T01:02:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.