{"about":{"site":"https://codewithpapers.app","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.","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"},"url":"/paper/improved-fine-tuning-of-large-multimodal-1","title":"Robust Adaptation of Large Multimodal Models for Retrieval Augmented Hateful Meme Detection","arxiv_id":"2502.13061","date":"2025-02-18","proceeding":null,"authors":["Jingbiao Mei","Jinghong Chen","Guangyu Yang","Weizhe Lin","Bill Byrne"],"abstract":"Hateful memes have become a significant concern on the Internet, necessitating robust automated detection systems. While LMMs have shown promise in hateful meme detection, they face notable challenges like sub-optimal performance and limited out-of-domain generalization capabilities. Recent studies further reveal the limitations of both SFT and in-context learning when applied to LMMs in this setting. To address these issues, we propose a robust adaptation framework for hateful meme detection that enhances in-domain accuracy and cross-domain generalization while preserving the general vision-language capabilities of LMMs. Experiments on six meme classification datasets show that our approach achieves state-of-the-art performance, outperforming larger agentic systems. Moreover, our method generates higher-quality rationales for explaining hateful content compared to standard SFT, enhancing model interpretability.","url_abs":"https://arxiv.org/abs/2502.13061v2","url_pdf":"https://arxiv.org/pdf/2502.13061v2.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":"improved-fine-tuning-of-large-multimodal-1","repo_url":"https://github.com/JingbiaoMei/RGCL","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"contrastive-learning","task_name":"Contrastive Learning"},{"task_slug":"domain-generalization","task_name":"Domain Generalization"},{"task_slug":"hateful-meme-classification","task_name":"Hateful Meme Classification"},{"task_slug":"in-context-learning","task_name":"In-Context Learning"},{"task_slug":"meme-classification","task_name":"Meme Classification"},{"task_slug":"retrieval","task_name":"Retrieval"}],"methods":[{"method_slug":"contrastive-learning","method_name":"Contrastive Learning"},{"method_slug":"sft","method_name":"SFT"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/hateful-meme-classification-on-harmeme","task":"Hateful Meme Classification","dataset":"HarMeme","model":"RA-HMD (Qwen2VL-7B)","rank_in_archive_order":1,"of":8,"metrics":{"AUROC":"93.2","Accuracy":"88.1"},"uses_additional_data":false},{"leaderboard":"/sota/hateful-meme-classification-on-harmeme","task":"Hateful Meme Classification","dataset":"HarMeme","model":"RA-HMD (Qwen2VL-2B)","rank_in_archive_order":2,"of":8,"metrics":{"AUROC":"92.9","Accuracy":"87.7"},"uses_additional_data":false},{"leaderboard":"/sota/hateful-meme-classification-on-harm-p","task":"Hateful Meme Classification","dataset":"Harm-P","model":"RA-HMD (Qwen2-VL-7B)","rank_in_archive_order":1,"of":5,"metrics":{"Accuracy":"91.6","F1":"91.1"},"uses_additional_data":false},{"leaderboard":"/sota/hateful-meme-classification-on-hateful-memes-1","task":"Hateful Meme Classification","dataset":"Hateful Memes","model":"RA-HMD (Qwen2-VL-7B)","rank_in_archive_order":1,"of":1,"metrics":{"AUROC":"91.1"},"uses_additional_data":false},{"leaderboard":"/sota/hateful-meme-classification-on-pridemm","task":"Hateful Meme Classification","dataset":"PrideMM","model":"RA-HMD (Qwen2-VL-7B)","rank_in_archive_order":1,"of":7,"metrics":{"Accuracy":"78.1","F1":"78.4"},"uses_additional_data":false},{"leaderboard":"/sota/hateful-meme-classification-on-pridemm","task":"Hateful Meme Classification","dataset":"PrideMM","model":"RA-HMD (Qwen2-VL-2B)","rank_in_archive_order":5,"of":7,"metrics":{"Accuracy":"76.0","F1":"76.7"},"uses_additional_data":false},{"leaderboard":"/sota/meme-classification-on-hateful-memes","task":"Meme Classification","dataset":"Hateful Memes","model":"RA-HMD (Qwen2-VL-7B)","rank_in_archive_order":1,"of":17,"metrics":{"Accuracy":"0.821","ROC-AUC":"0.911"},"uses_additional_data":false},{"leaderboard":"/sota/meme-classification-on-hateful-memes","task":"Meme Classification","dataset":"Hateful Memes","model":"RA-HMD (LLaVA-1.5-7B)","rank_in_archive_order":2,"of":17,"metrics":{"Accuracy":"0.809","ROC-AUC":"0.897"},"uses_additional_data":false},{"leaderboard":"/sota/meme-classification-on-hateful-memes","task":"Meme Classification","dataset":"Hateful Memes","model":"RA-HMD (Qwen2-VL-2B)","rank_in_archive_order":4,"of":17,"metrics":{"Accuracy":"0.791","ROC-AUC":"0.884"},"uses_additional_data":false},{"leaderboard":"/sota/meme-classification-on-multioff","task":"Meme Classification","dataset":"MultiOFF","model":"RA-HMD (Qwen2-VL-7B)","rank_in_archive_order":1,"of":5,"metrics":{"Accuracy":"71.1","F1":"64.8"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2502.13061","atlas_url":"https://app.syntology.ai/?focus=2502.13061","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2502.13061"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"deterministic:regex_extraction","url":"https://github.com/JingbiaoMei/RGCL","reach":null}],"summary":{"ran_fixture":1,"ran_draft_wrong":1,"unverified":2},"by_repo_kind":{"official":{"samples":4,"ran":2,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"b79d1b8058e01331","entry":"compute_l2","repo":"JingbiaoMei/RGCL","repo_kind":"official","path":"src/model/loss.py","file_url":"https://github.com/JingbiaoMei/RGCL/blob/HEAD/src/model/loss.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"b79d1b8058e01331"}},{"code_sha256_prefix":"a20178fd77f8b5a5","entry":"sparse_retrieve_hard_negatives_pseudo_positive","repo":"JingbiaoMei/RGCL","repo_kind":"official","path":"src/model/loss.py","file_url":"https://github.com/JingbiaoMei/RGCL/blob/HEAD/src/model/loss.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a20178fd77f8b5a5"}},{"code_sha256_prefix":"e29c3311f99491d1","entry":"compute_loss","repo":"JingbiaoMei/RGCL","repo_kind":"official","path":"src/model/loss.py","file_url":"https://github.com/JingbiaoMei/RGCL/blob/HEAD/src/model/loss.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e29c3311f99491d1"}},{"code_sha256_prefix":"20d79017dbff7d0c","entry":"dense_retrieve_hard_negatives_pseudo_positive","repo":"JingbiaoMei/RGCL","repo_kind":"official","path":"src/model/loss.py","file_url":"https://github.com/JingbiaoMei/RGCL/blob/HEAD/src/model/loss.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"20d79017dbff7d0c"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}