Papers › CoCoT: Contrastive Chain-of-Thought Prompting for Large Multimodal Models with...

CoCoT: Contrastive Chain-of-Thought Prompting for Large Multimodal Models with Multiple Image Inputs

5 Jan 2024arXiv:2401.02582archive 2025-07-28

Daoan Zhang, Junming Yang, Hanjia Lyu, Zijian Jin, Yuan YAO, Mingkai Chen, Jiebo Luo

When exploring the development of Artificial General Intelligence (AGI), a critical task for these models involves interpreting and processing information from multiple image inputs. However, Large Multimodal Models (LMMs) encounter two issues in such scenarios: (1) a lack of fine-grained perception, and (2) a tendency to blend information across multiple images. We first extensively investigate the capability of LMMs to perceive fine-grained visual details when dealing with multiple input images. The research focuses on two aspects: first, image-to-image matching (to evaluate whether LMMs can effectively reason and pair relevant images), and second, multi-image-to-text matching (to assess whether LMMs can accurately capture and summarize detailed image information). We conduct evaluations on a range of both open-source and closed-source large models, including GPT-4V, Gemini, OpenFlamingo, and MMICL. To enhance model performance, we further develop a Contrastive Chain-of-Thought (CoCoT) prompting approach based on multi-input multimodal models. This method requires LMMs to compare the similarities and differences among multiple image inputs, and then guide the models to answer detailed questions about multi-image inputs based on the identified similarities and differences. Our experimental results showcase CoCoT's proficiency in enhancing the multi-image comprehension capabilities of large multimodal models.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

vista-h/gpt-4v_social_media mentioned on GitHub report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image ComprehensionImage to textText MatchingVisual Reasoning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Visual Reasoning Winoground MMICL + CoCoT Group Score 50.75 #4 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + CoCoT Image Score 52.5 #4 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + CoCoT Text Score 64.25 #4 of 114 Archive leaderboard report
Visual Reasoning Winoground GPT-4V + CoCoT Group Score 44.5 #5 of 114 Archive leaderboard report
Visual Reasoning Winoground GPT-4V + CoCoT Image Score 49.5 #5 of 114 Archive leaderboard report
Visual Reasoning Winoground GPT-4V + CoCoT Text Score 58.5 #5 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + CoCoT Group Score 41.5 #6 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + CoCoT Image Score 55.25 #6 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + CoCoT Text Score 58.25 #6 of 114 Archive leaderboard report
Visual Reasoning Winoground GPT-4V Group Score 37.75 #7 of 114 Archive leaderboard report
Visual Reasoning Winoground GPT-4V Image Score 42.5 #7 of 114 Archive leaderboard report
Visual Reasoning Winoground GPT-4V Text Score 54.5 #7 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + CCoT Group Score 47.5 #10 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + CCoT Image Score 48 #10 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + CCoT Text Score 51 #10 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + DDCoT Group Score 39 #11 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + DDCoT Image Score 47.25 #11 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + DDCoT Text Score 47.5 #11 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + DDCoT Group Score 36.75 #13 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + DDCoT Image Score 45 #13 of 114 Archive leaderboard report
Visual Reasoning Winoground MMICL + DDCoT Text Score 46.75 #13 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + DDCoT Group Score 23.75 #19 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + DDCoT Image Score 25 #19 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + DDCoT Text Score 45 #19 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + CCoT Group Score 20 #28 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + CCoT Image Score 27.5 #28 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo + CCoT Text Score 42.5 #28 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + CoCoT Group Score 27.75 #37 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + CoCoT Image Score 32.5 #37 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + CoCoT Text Score 40 #37 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo Group Score 33.25 #40 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo Image Score 41.25 #40 of 114 Archive leaderboard report
Visual Reasoning Winoground OpenFlamingo Text Score 39 #40 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini Group Score 25 #63 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini Image Score 26 #63 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini Text Score 30.75 #63 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + CCoT Group Score 20.75 #96 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + CCoT Image Score 33 #96 of 114 Archive leaderboard report
Visual Reasoning Winoground Gemini + CCoT Text Score 22.5 #96 of 114 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections