{"url":"/dataset/benchlmm","name":"BenchLMM","full_name":"BenchLMM: Benchmarking Cross-style Visual Capability of Large Multimodal Models","description_markdown":"Large Multimodal Models (LMMs) such as GPT-4V and LLaVA have shown remarkable capabilities in visual reasoning with common image styles. However, their robustness against diverse style shifts, crucial for practical applications, remains largely unexplored. In this paper, we propose a new benchmark, BenchLMM, to assess the robustness of LMMs against three different styles: artistic image style, imaging sensor style, and application style, where each style has five sub-styles. Utilizing BenchLMM, we comprehensively evaluate state-of-the-art LMMs and reveal: 1) LMMs generally suffer performance degradation when working with other styles; 2) An LMM performs better than another model in common style does not guarantee its superior performance in other styles; 3) LMMs' reasoning capability can be enhanced by prompting LMMs to predict the style first, based on which we propose a versatile and training-free method for improving LMMs; 4) An intelligent LMM is expected to interpret the causes of its errors when facing stylistic variations. We hope that our benchmark and analysis can shed new light on developing more intelligent and versatile LMMs.","description_withheld":null,"homepage":"https://aifeg.github.io/BenchLMM/","introduced_date":"2023-12-05","introduced_date_note":null,"introduced_by":{"paper":"/paper/benchlmm-benchmarking-cross-style-visual","title":"BenchLMM: Benchmarking Cross-style Visual Capability of Large Multimodal Models","first_author":"Rizhao Cai","url":null},"license":{"name":"Apache 2.0","url":"https://github.com/AIFEG/BenchLMM/blob/main/LICENSE"},"modalities":[{"name":"Images","url":"/datasets/modality/images"},{"name":"Texts","url":"/datasets/modality/texts"}],"tasks":[{"name":"Visual Question Answering (VQA)","url":"/task/visual-question-answering","datasets_with_task":"/datasets/task/visual-question-answering"},{"name":"Visual Question Answering","url":"/task/visual-question-answering-1","datasets_with_task":"/datasets/task/visual-question-answering-1"}],"languages":[{"name":"English","url":"/datasets/language/english"}],"variants":["BenchLMM"],"data_loaders":[],"num_papers_in_archive":12,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/visual-question-answering-on-benchlmm","task":"Visual Question Answering","dataset_variant":"BenchLMM","rows":10,"metrics":["GPT-3.5 score"],"first_row_in_archive_order":{"model":"GPT-4V","paper":"/paper/gpt-4-technical-report-1","metrics":{"GPT-3.5 score":"58.37"},"code_links":[{"title":"openai/evals","url":"https://github.com/openai/evals"},{"title":"shmsw25/factscore","url":"https://github.com/shmsw25/factscore"},{"title":"unispac/visual-adversarial-examples-jailbreak-large-language-models","url":"https://github.com/unispac/visual-adversarial-examples-jailbreak-large-language-models"},{"title":"gpt4life/alpagasus","url":"https://github.com/gpt4life/alpagasus"},{"title":"emrgnt-cmplxty/zero-shot-replication","url":"https://github.com/emrgnt-cmplxty/zero-shot-replication"},{"title":"ethz-privsec/superhuman-ai-consistency","url":"https://github.com/ethz-privsec/superhuman-ai-consistency"},{"title":"ethz-spylab/superhuman-ai-consistency","url":"https://github.com/ethz-spylab/superhuman-ai-consistency"},{"title":"eternityyw/tram-benchmark","url":"https://github.com/eternityyw/tram-benchmark"},{"title":"AUCOHL/RTL-Repo","url":"https://github.com/AUCOHL/RTL-Repo"},{"title":"zach-zhiling-zheng/reticular_chemist","url":"https://github.com/zach-zhiling-zheng/reticular_chemist"},{"title":"lflage/openfactscore","url":"https://github.com/lflage/openfactscore"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/sphinx-the-joint-mixing-of-weights-tasks-and","title":"SPHINX: The Joint Mixing of Weights, Tasks, and Visual Embeddings for Multi-modal Large Language Models","date":"2023-11-13","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":3,"samples_ran":3,"samples_unverified":0,"pointer_only_for_licence":3,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/minigpt-v2-large-language-model-as-a-unified","title":"MiniGPT-v2: large language model as a unified interface for vision-language multi-task learning","date":"2023-10-14","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":3,"samples_ran":3,"samples_unverified":0,"pointer_only_for_licence":1,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/improved-baselines-with-visual-instruction","title":"Improved Baselines with Visual Instruction Tuning","date":"2023-10-05","rows_on_this_dataset":1,"code_links":9,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":9,"samples_ran":6,"samples_unverified":3,"pointer_only_for_licence":8,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/instructblip-towards-general-purpose-vision","title":"InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning","date":"2023-05-11","rows_on_this_dataset":2,"code_links":4,"syntology":null},{"paper":"/paper/otter-a-multi-modal-model-with-in-context","title":"Otter: A Multi-Modal Model with In-Context Instruction Tuning","date":"2023-05-05","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/minigpt-4-enhancing-vision-language","title":"MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models","date":"2023-04-20","rows_on_this_dataset":1,"code_links":6,"syntology":null},{"paper":"/paper/visual-instruction-tuning-1","title":"Visual Instruction Tuning","date":"2023-04-17","rows_on_this_dataset":2,"code_links":13,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":51,"samples_ran":16,"samples_unverified":35,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/gpt-4-technical-report-1","title":"GPT-4 Technical Report","date":"2023-03-15","rows_on_this_dataset":1,"code_links":11,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":5,"samples_ran":2,"samples_unverified":3,"pointer_only_for_licence":1,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}}],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":5,"samples_harvested":71,"samples_ran":30,"samples_unverified":41,"pointer_only_for_licence":13,"papers_with_no_sample_that_ran":0,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}