{"about":{"non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","site":"https://codewithpapers.app","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page","syntology":{"site":"https://syntology.ai","developers":"https://syntology.ai/developers","mcp":{"server":"https://syntology.ai/mcp","transport":"streamable-http","server_card":"https://syntology.ai/.well-known/mcp/server-card.json","auth":{"type":"trial token, no account","trial_token":"https://syntology.ai/api/oauth/trial/token","method":"POST","docs":"https://syntology.ai/developers"}},"have":"https://syntology.ai/api/graph/have?x=<method, arXiv id or title> (free, answers coverage only)","paper_base":"https://syntology.ai/paper/","atlas_base":"https://app.syntology.ai/?focus="},"machine_readable":[{"url":"https://codewithpapers.app/llms.txt","what":"the machine catalog: every machine-readable file, counted"},{"url":"https://codewithpapers.app/index/manifest.json","what":"paper-to-code index by arXiv id, with Syntology's counts"},{"url":"https://codewithpapers.app/search/manifest.json","what":"site search index (titles, authors) and its files"},{"url":"https://codewithpapers.app/download","what":"bulk files: Syntology's layer, described there"},{"url":"https://codewithpapers.app/build_manifest.json","what":"the build record: inputs, counts, exclusions, probes"}]},"url":"/paper/rlhf-v-towards-trustworthy-mllms-via-behavior","title":"RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback","arxiv_id":"2312.00849","date":"2023-12-01","proceeding":"CVPR 2024 1","authors":["Tianyu Yu","Yuan YAO","Haoye Zhang","Taiwen He","Yifeng Han","Ganqu Cui","Jinyi Hu","Zhiyuan Liu","Hai-Tao Zheng","Maosong Sun","Tat-Seng Chua"],"abstract":"Multimodal Large Language Models (MLLMs) have recently demonstrated impressive capabilities in multimodal understanding, reasoning, and interaction. However, existing MLLMs prevalently suffer from serious hallucination problems, generating text that is not factually grounded in associated images. The problem makes existing MLLMs untrustworthy and thus impractical in real-world (especially high-stakes) applications. To address the challenge, we present RLHF-V, which enhances MLLM trustworthiness via behavior alignment from fine-grained correctional human feedback. Specifically, RLHF-V collects human preference in the form of segment-level corrections on hallucinations, and performs dense direct preference optimization over the human feedback. Comprehensive experiments on five benchmarks in both automatic and human evaluation show that, RLHF-V can enable substantially more trustworthy MLLM behaviors with promising data and computation efficiency. Remarkably, using 1.4k annotated data samples, RLHF-V significantly reduces the hallucination rate of the base MLLM by 34.8%, outperforming the concurrent LLaVA-RLHF trained on 10k annotated data. The final model achieves state-of-the-art performance in trustworthiness among open-source MLLMs, and shows better robustness than GPT-4V in preventing hallucinations aroused from over-generalization. We open-source our code, model, and data at https://github.com/RLHF-V/RLHF-V.","url_abs":"https://arxiv.org/abs/2312.00849v2","url_pdf":"https://arxiv.org/pdf/2312.00849v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"rlhf-v-towards-trustworthy-mllms-via-behavior","repo_url":"https://github.com/rlhf-v/rlhf-v","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"rlhf-v-towards-trustworthy-mllms-via-behavior","repo_url":"https://github.com/exgc/r1v-free","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"rlhf-v-towards-trustworthy-mllms-via-behavior","repo_url":"https://github.com/openbmb/minicpm-v","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"rlhf-v-towards-trustworthy-mllms-via-behavior","repo_url":"https://github.com/tidedra/vl-rlhf","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"hallucination","task_name":"Hallucination"},{"task_slug":"image-captioning","task_name":"Image Captioning"},{"task_slug":"visual-question-answering-1","task_name":"Visual Question Answering"}],"methods":[{"method_slug":"base","method_name":"BASE"}],"datasets_introduced":[{"slug":"rlhf-v-dataset","name":"RLHF-V Dataset","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-captioning-on-object-halbench","task":"Image Captioning","dataset":"Object HalBench","model":"RLHF-V","rank_in_archive_order":1,"of":3,"metrics":{"chair_i":"7.5","chair_s":"12.2"},"uses_additional_data":false},{"leaderboard":"/sota/visual-question-answering-on-vqa-v2-1","task":"Visual Question Answering","dataset":"VQA v2","model":"RLHF-V","rank_in_archive_order":1,"of":2,"metrics":{"Accuracy":"80"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2312.00849","atlas_url":"https://app.syntology.ai/?focus=2312.00849","mcp":null,"developers":"https://syntology.ai/developers","held":true,"reason":"arxiv_id_in_atlas","checked_against":{"atlas_ids":549449,"atlas_ids_sha256":"afd72399dafd9906e9062927f70655952385a3d75488558e1f0484e950ab09b7","read_at":"2026-10-05T01:46:21Z"},"agent_calls":[{"tool":"get_citation_path","arguments":{"paper_1":"2312.00849"},"arguments_in":{"paper_2":"another paper's arXiv id or title"},"call":"get_citation_path(paper_1=\"2312.00849\", paper_2=\"…\")"},{"tool":"get_concepts_for_paper","arguments":{"arxiv_id":"2312.00849"},"call":"get_concepts_for_paper(arxiv_id=\"2312.00849\")"}],"by_repo":[{"repo":"rlhf-v/rlhf-v","kind":"official","samples":0,"ran":0,"instrument":0,"unverified":0,"harvested_for_other_papers":1,"harvested_for":["2405.17220"],"harvested_for_calls":[{"arxiv_id":"2405.17220","tool":"get_harvested_code_for_paper","arguments":{"arxiv_id":"2405.17220"},"call":"get_harvested_code_for_paper(arxiv_id=\"2405.17220\")"}]},{"repo":"exgc/r1v-free","kind":"listed","samples":0,"ran":0,"instrument":0,"unverified":0},{"repo":"openbmb/minicpm-v","kind":"listed","samples":0,"ran":0,"instrument":0,"unverified":0,"harvested_for_other_papers":1,"harvested_for":["2408.01800"],"harvested_for_calls":[{"arxiv_id":"2408.01800","tool":"get_harvested_code_for_paper","arguments":{"arxiv_id":"2408.01800"},"call":"get_harvested_code_for_paper(arxiv_id=\"2408.01800\")"}]},{"repo":"tidedra/vl-rlhf","kind":"listed","samples":0,"ran":0,"instrument":0,"unverified":0,"harvested_for_other_papers":1,"harvested_for":["2025.naacl-long.447"],"harvested_for_calls":[]}],"by_repo_harvested_for_is":"harvested_for: the arXiv ids (as Syntology's graph holds them) of every paper whose samples Syntology's graph links to this repository, in arXiv-id order, complete; the page names the first 3 and says how many more. A link only: it does not say the papers are the same work, that one copied the other, or that the code implements this paper's method; harvested_for_calls gives the get_harvested_code_for_paper call for each arXiv-shaped id among them","by_repo_is":"one row per repository in the archive's code links, Syntology's graph links or the samples; samples 0 means no sample is linked to this paper, a statement about Syntology's coverage, not about the repository: with harvested_for_other_papers present, Syntology harvested the repository for that many other papers, and without it, Syntology harvested nothing from it; placed_by_identical_code counts samples whose link records no repository, placed here because identical code was harvested from this repository for another paper; ran counts executions on a synthesized input and instrument counts failures of Syntology's instrument, not of the code"},"code_links_note":null,"arxiv_metadata":null,"syntology_extracted_results":null,"record_sha256":"41f8136db7df3bc0f4feac969ad3eeaec699656b7187bed48be579688a2fada5","record_changed_at":"2026-09-28","record_changed_at_basis":"first_hashed"}