Beyond Multiple Choice: Verifiable OpenQA for Robust Vision-Language RFT

• URL: http://arxiv.org/abs/2511.17405v2
• Date: Mon, 24 Nov 2025 02:45:02 GMT
• Title: Beyond Multiple Choice: Verifiable OpenQA for Robust Vision-Language RFT
• Abstract summary: Multiple-choice question answering (MCQA) has been a popular format for evaluating and reinforcement fine-tuning (RFT) of modern multimodal language models.<n>We propose ReVeL, a framework that rewrites multiple-choice questions into open-form questions while keeping answers verifiable whenever possible.
• Score: 25.508458982314284
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Multiple-choice question answering (MCQA) has been a popular format for evaluating and reinforcement fine-tuning (RFT) of modern multimodal language models. Its constrained output format allows for simplified, deterministic automatic verification. However, we find that the options may leak exploitable signals, which makes the accuracy metrics unreliable for indicating real capabilities and encourages explicit or implicit answer guessing behaviors during RFT. We propose ReVeL (Rewrite and Verify by LLM), a framework that rewrites multiple-choice questions into open-form questions while keeping answers verifiable whenever possible. The framework categorizes questions according to different answer types, apply different rewriting and verification schemes, respectively. When applied for RFT, we converted 20k MCQA examples and use GRPO to finetune Qwen2.5-VL models. Models trained on ReVeL-OpenQA match MCQA accuracy on multiple-choice benchmarks and improve OpenQA accuracy by about six percentage points, indicating better data efficiency and more robust reward signals than MCQA-based training. When used for evaluation, ReVeL also reveals up to 20 percentage points of score inflation in MCQA benchmarks (relative to OpenQA), improves judging accuracy, and reduces both cost and latency. We will release code and data publicly.

Related papers

• Check The Scoreboard: An Analysis of Scoring Schemes on Multiple-Choice Evaluation [52.58737865652009]
  Multiple-choice question answering (MCQA) benchmarks in NLP use number-right scoring (accuracy)<n>In educational testing, the scoring scheme, the combination of the response mode models follow and the rule for grading responses, is a key design choice that dictates which abilities to reward.<n>We examine how alternatives to number right change what MCQA measures with six education-inspired schemes that assess abilities beyond accuracy.
  arXiv  Detail & Related papers  (2026-08-30T16:35:22Z)
• MinosEval: Distinguishing Factoid and Non-Factoid for Tailored Open-Ended QA Evaluation with LLMs [15.278241998033822]
  Open-ended question answering (QA) is a key task for evaluating the capabilities of large language models (LLMs)<n>We propose textbfMinosEval, a novel evaluation method that first distinguishes open-ended questions and then ranks candidate answers.
  arXiv  Detail & Related papers  (2025-06-18T07:49:13Z)
• Trust but Verify: Programmatic VLM Evaluation in the Wild [62.14071929143684]
  Programmatic VLM Evaluation (PROVE) is a new benchmarking paradigm for evaluating VLM responses to open-ended queries. We benchmark the helpfulness-truthfulness trade-offs of a range ofVLMs on PROVE, finding that very few are in-fact able to achieve a good balance between the two.
  arXiv  Detail & Related papers  (2024-10-17T01:19:18Z)
• Listening to the Wise Few: Select-and-Copy Attention Heads for Multiple-Choice QA [19.78468832417275]
  We introduce new scores that better capture and reveal model's underlying knowledge. Based on these scores, our method improves knowledge extraction, yielding up to 16% gain for LLaMA2-7B. The accuracy on a simple synthetic dataset, where the model explicitly knows the right answer, increases by almost 60%.
  arXiv  Detail & Related papers  (2024-10-03T09:53:48Z)
• Answer, Assemble, Ace: Understanding How LMs Answer Multiple Choice Questions [103.20281438405111]
  Multiple-choice question answering (MCQA) is a key competence of performant transformer language models.<n>We employ vocabulary projection and activation patching methods to localize key hidden states that encode relevant information for predicting the correct answer.<n>We show that subsequent layers increase the probability of the predicted answer symbol in vocabulary space, and that this probability increase is associated with a sparse set of attention heads with unique roles.
  arXiv  Detail & Related papers  (2024-07-21T00:10:23Z)
• MinPrompt: Graph-based Minimal Prompt Data Augmentation for Few-shot Question Answering [64.6741991162092]
  We present MinPrompt, a minimal data augmentation framework for open-domain question answering. We transform the raw text into a graph structure to build connections between different factual sentences. We then apply graph algorithms to identify the minimal set of sentences needed to cover the most information in the raw text. We generate QA pairs based on the identified sentence subset and train the model on the selected sentences to obtain the final model.
  arXiv  Detail & Related papers  (2023-10-08T04:44:36Z)
• SQUARE: Automatic Question Answering Evaluation using Multiple Positive and Negative References [73.67707138779245]
  We propose a new evaluation metric: SQuArE (Sentence-level QUestion AnsweRing Evaluation) We evaluate SQuArE on both sentence-level extractive (Answer Selection) and generative (GenQA) QA systems.
  arXiv  Detail & Related papers  (2023-09-21T16:51:30Z)
• Open-vocabulary Video Question Answering: A New Benchmark for Evaluating the Generalizability of Video Question Answering Models [15.994664381976984]
  We introduce a new benchmark, Open-vocabulary Video Question Answering (OVQA), to measure the generalizability of VideoQA models. In addition, we introduce a novel GNN-based soft verbalizer that enhances the prediction on rare and unseen answers. Our ablation studies and qualitative analyses demonstrate that our GNN-based soft verbalizer further improves the model performance.
  arXiv  Detail & Related papers  (2023-08-18T07:45:10Z)
• RoMQA: A Benchmark for Robust, Multi-evidence, Multi-answer Question Answering [87.18962441714976]
  We introduce RoMQA, the first benchmark for robust, multi-evidence, multi-answer question answering (QA) We evaluate state-of-the-art large language models in zero-shot, few-shot, and fine-tuning settings, and find that RoMQA is challenging. Our results show that RoMQA is a challenging benchmark for large language models, and provides a quantifiable test to build more robust QA methods.
  arXiv  Detail & Related papers  (2022-10-25T21:39:36Z)
• Few-shot Reranking for Multi-hop QA via Language Model Prompting [56.454088569241534]
  We study few-shot reranking for multi-hop QA with open-domain questions. We propose PromptRank, which relies on large language models prompting for multi-hop path reranking. PromptRank yields strong retrieval performance on HotpotQA with only 128 training examples.
  arXiv  Detail & Related papers  (2022-05-25T10:45:55Z)
• Mitigating False-Negative Contexts in Multi-document QuestionAnswering with Retrieval Marginalization [29.797379277423143]
  We develop a new parameterization of set-valued retrieval that properly handles unanswerable queries. We show that marginalizing over this set during training allows a model to mitigate false negatives in annotated supporting evidences. On IIRC, we show that joint modeling with marginalization on alternative contexts improves model performance by 5.5 F1 points and achieves a new state-of-the-art performance of 50.6 F1.
  arXiv  Detail & Related papers  (2021-03-22T23:44:35Z)
• Counterfactual Variable Control for Robust and Interpretable Question Answering [57.25261576239862]
  Deep neural network based question answering (QA) models are neither robust nor explainable in many cases. In this paper, we inspect such spurious "capability" of QA models using causal inference. We propose a novel approach called Counterfactual Variable Control (CVC) that explicitly mitigates any shortcut correlation.
  arXiv  Detail & Related papers  (2020-10-12T10:09:05Z)
• Selective Question Answering under Domain Shift [90.021577320085]
  Abstention policies based solely on the model's softmax probabilities fare poorly, since models are overconfident on out-of-domain inputs. We train a calibrator to identify inputs on which the QA model errs, and abstain when it predicts an error is likely. Our method answers 56% of questions while maintaining 80% accuracy; in contrast, directly using the model's probabilities only answers 48% at 80% accuracy.
  arXiv  Detail & Related papers  (2020-06-16T19:13:21Z)

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.