mlsanityData to 7 Oct 2026

Unsolvable Problem Detection: Robust Understanding Evaluation for Large Multimodal Models

arXiv 2403.20331 ↗First posted 29 Mar 2024cs.CV cs.AI cs.CLcited by 12

On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning

Abstract (arXiv)

This paper introduces a novel task to evaluate the robust understanding capability of Large Multimodal Models (LMMs), termed Unsolvable Problem Detection (UPD). Multiple-choice question answering (MCQA) is widely used to assess the understanding capability of LMMs, but it does not guarantee that LMMs truly comprehend the answer. UPD assesses the LMM's ability to withhold answers when encountering unsolvable problems of MCQA, verifying whether the model truly understands the answer. UPD encompasses three problems: Absent Answer Detection (AAD), Incompatible Answer Set Detection (IASD), and Incompatible Visual Question Detection (IVQD), covering unsolvable cases like answer-lacking or incompatible choices and image-question mismatches. For the evaluation, we introduce the MM-UPD Bench, a benchmark for assessing performance across various ability dimensions. Our experiments reveal that even most LMMs, which demonstrate adequate performance on existing benchmarks, struggle significantly with MM-UPD, underscoring a novel aspect of trustworthiness that current benchmarks have overlooked. A detailed analysis shows that LMMs have different bottlenecks and chain-of-thought and self-reflection improved performance for LMMs with the bottleneck in their LLM capability. We hope our insights will enhance the broader understanding and development of more reliable LMMs. The code is available at https://github.com/AtsuMiyai/UPD.

Key facts. Cited by 12 papers in the map (data through 7 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 59 of its 95 references that are in the map. First posted on arXiv 29 Mar 2024.

Builds on

59 of 95 references.

Closest ancestors · 8

Same direction, newest first

  1. 2023VILA: On Pre-training for Visual Language Models12 Dec
  2. 2023mPLUG-Owl2: Revolutionizing Multi-modal Large Language Model with Modality Collaboration7 Nov
  3. 2023Improved Baselines with Visual Instruction Tuning5 Oct
  4. 2023Aligning Large Multimodal Models with Factually Augmented RLHF25 Sep
  5. 2023MMICL: Empowering Vision-language Model with Multi-Modal In-Context Learning14 Sep
  6. 2023MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities4 Aug
  7. 2023InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning11 May
  8. 2023Otter: A Multi-Modal Model with In-Context Instruction Tuning5 May

Other references · 51

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  3. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  4. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,337
  5. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  6. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,175
  7. 2016Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answeringcited by 2,498
  8. 2015VQA: Visual Question Answeringcited by 2,489
  9. 2016A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networkscited by 2,362
  10. 2023MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGIcited by 2,044
  11. 2022Language Models (Mostly) Know What They Knowcited by 1,783
  12. 2016Concrete Problems in AI Safetycited by 1,582
38 more
  1. 2023MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contextscited by 1,582
  2. 2018Know What You Don't Know: Unanswerable Questions for SQuADcited by 1,538
  3. 2017Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networkscited by 1,325
  4. 2020Oscar: Object-Semantics Aligned Pre-training for Vision-Language Taskscited by 1,213
  5. 2019OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledgecited by 1,114
  6. 2023Large Language Models are not Fair Evaluatorscited by 1,050
  7. 2023mPLUG-Owl: Modularization Empowers Large Language Models with Multimodalitycited by 1,038
  8. 2018VizWiz Grand Challenge: Answering Visual Questions from Blind Peoplecited by 852
  9. 2021Generalized Out-of-Distribution Detection: A Surveycited by 796
  10. 2023LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attentioncited by 776
  11. 2022A-OKVQA: A Benchmark for Visual Question Answering using World Knowledgecited by 702
  12. 2018CoQA: A Conversational Question Answering Challengecited by 674
  13. 2023LLaMA-Adapter V2: Parameter-Efficient Visual Instruction Modelcited by 602
  14. 2024MathVerse: Does Your Multi-modal LLM Truly See the Diagrams in Visual Math Problems?cited by 573
  15. 2019Unified Vision-Language Pre-Training for Image Captioning and VQAcited by 494
  16. 2023OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocationcited by 468
  17. 2023Large Language Models Are Not Robust Multiple Choice Selectorscited by 454
  18. 2023Larger language models do in-context learning differentlycited by 386
  19. 2018QuAC : Question Answering in Contextcited by 379
  20. 2022VOS: Learning What You Don't Know by Virtual Outlier Synthesiscited by 300
  21. 2021Unsolved Problems in ML Safetycited by 291
  22. 2022Delving into Out-of-Distribution Detection with Vision-Language Representationscited by 215
  23. 2022Leveraging Large Language Models for Multiple Choice Question Answeringcited by 199
  24. 2024Mitigating Hallucinations in Large Vision-Language Models with Instruction Contrastive Decodingcited by 198
  25. 2024Don't Hallucinate, Abstain: Identifying LLM Knowledge Gaps via Multi-LLM Collaborationcited by 176
  26. 2023SVIT: Scaling up Visual Instruction Tuningcited by 161
  27. 2023Holistic Analysis of Hallucination in GPT-4V(ision): Bias and Interference Challengescited by 131
  28. 2023CLIPN for Zero-Shot OOD Detection: Teaching CLIP to Say Nocited by 130
  29. 2019Unsupervised Out-of-Distribution Detection by Maximum Classifier Discrepancycited by 112
  30. 2023LoCoOp: Few-Shot Out-of-Distribution Detection via Prompt Learningcited by 107
  31. 2021Can multi-label classification networks know what they don't know?cited by 100
  32. 2022Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectlycited by 86
  33. 2022X-Risk Analysis for AI Researchcited by 60
  34. 2019Why Does a Visual Question Have Different Answers?cited by 34
  35. 2024LLMs May Perform MCQA by Selecting the Least Incorrect Optioncited by 32
  36. 2023UNK-VQA: A Dataset and a Probe into the Abstention Ability of Multi-modal Large Modelscited by 23
  37. 2024VISREAS: Complex Visual Reasoning with Unanswerable Questionscited by 9
  38. 2021Unanswerable Questions about Images and Textscited by 7

50 most cited of 51 All 51 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 4; 2025: 3; 2026: 5.

12 papers.

  1. 2024Is A Picture Worth A Thousand Words? Delving Into Spatial Reasoning for Vision Language Modelscited by 111
  2. 2024Generalized Out-of-Distribution Detection and Beyond in Vision Language Model Era: A Surveycited by 36
  3. 2024AutoTrust: Benchmarking Trustworthiness in Large Vision Language Models for Autonomous Drivingcited by 35
  4. 2024On the Learnability of Out-of-distribution Detectioncited by 8
  5. 2025CLIP-UP: CLIP-Based Unanswerable Problem Detection for Visual Question Answeringcited by 6
  6. 2026AQUA-Bench: Beyond Finding Answers to Knowing When There Are None in Audio Question Answeringcited by 6
  7. 2026TRAPSBench: Vision-Language Models Encode but Fail to Express Epistemic Restraintcited by 3
  8. 2025CLASH: A Benchmark for Cross-Modal Contradiction Detectioncited by 2
  9. 2025JMMMU-Pro: Image-based Japanese Multi-discipline Multimodal Understanding Benchmark via Vibe Benchmark Constructioncited by 0
  10. 2026When No Answer Is Correct: Diagnosing Absent Answer Detection for MLLMs in Video Understandingcited by 0
  11. 2026MMOOC: A Comprehensive Benchmark for Out-of-Context Evaluation in Multimodal Large Language Modelscited by 0
  12. 2026From Knowing to Abstaining: Bridging the Representation-Action Gap in Vision-Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024MMEvalPro: Calibrating Multimodal Benchmarks Towards Trustworthy and Efficient Evaluation
  2. 2026MetaRA: Metamorphic Robustness Assessment for Multimodal Large Language Model-based Visual Question Answering Systems
  3. 2024Exploring Response Uncertainty in MLLMs: An Empirical Evaluation under Misleading Scenarios
  4. 2024Ranked from Within: Ranking Large Multimodal Models Without Labels
  5. 2026Knowing When Not to Answer: Evaluating Abstention in Multimodal Reasoning Systems
  6. 2026Seeing and Solving Are Not Enough for Vision-Language Models
  7. 2026Evaluating Cross-Modal Reasoning Ability and Problem Characteristics with Multimodal Item Response Theory
  8. 2025UPME: An Unsupervised Peer Review Framework for Multimodal Large Language Model Evaluation