{"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/llava-med-training-a-large-language-and","title":"LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day","arxiv_id":"2306.00890","date":"2023-06-01","proceeding":"NeurIPS 2023 11","authors":["Chunyuan Li","Cliff Wong","Sheng Zhang","Naoto Usuyama","Haotian Liu","Jianwei Yang","Tristan Naumann","Hoifung Poon","Jianfeng Gao"],"abstract":"Conversational generative AI has demonstrated remarkable promise for empowering biomedical practitioners, but current investigations focus on unimodal text. Multimodal conversational AI has seen rapid progress by leveraging billions of image-text pairs from the public web, but such general-domain vision-language models still lack sophistication in understanding and conversing about biomedical images. In this paper, we propose a cost-efficient approach for training a vision-language conversational assistant that can answer open-ended research questions of biomedical images. The key idea is to leverage a large-scale, broad-coverage biomedical figure-caption dataset extracted from PubMed Central, use GPT-4 to self-instruct open-ended instruction-following data from the captions, and then fine-tune a large general-domain vision-language model using a novel curriculum learning method. Specifically, the model first learns to align biomedical vocabulary using the figure-caption pairs as is, then learns to master open-ended conversational semantics using GPT-4 generated instruction-following data, broadly mimicking how a layperson gradually acquires biomedical knowledge. This enables us to train a Large Language and Vision Assistant for BioMedicine (LLaVA-Med) in less than 15 hours (with eight A100s). LLaVA-Med exhibits excellent multimodal conversational capability and can follow open-ended instruction to assist with inquiries about a biomedical image. On three standard biomedical visual question answering datasets, LLaVA-Med outperforms previous supervised state-of-the-art on certain metrics. To facilitate biomedical multimodal research, we will release our instruction-following data and the LLaVA-Med model.","url_abs":"https://arxiv.org/abs/2306.00890v1","url_pdf":"https://arxiv.org/pdf/2306.00890v1.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":"llava-med-training-a-large-language-and","repo_url":"https://github.com/microsoft/LLaVA-Med","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"instruction-following","task_name":"Instruction Following"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"referring-expression-comprehension","task_name":"Referring Expression Comprehension"},{"task_slug":"referring-expression-generation","task_name":"Referring expression generation"},{"task_slug":"visual-question-answering-1","task_name":"Visual Question Answering"}],"methods":[{"method_slug":"align","method_name":"ALIGN"},{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"focus","method_name":"Focus"},{"method_slug":"gpt-4","method_name":"GPT-4"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-classification-on-coloninst-v1-seen","task":"Image Classification","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.0 (w/o LoRA, w/ extra data)","rank_in_archive_order":2,"of":17,"metrics":{"Accuray":"93.84"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-coloninst-v1-seen","task":"Image Classification","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.5 (w/ LoRA, w/o extra data)","rank_in_archive_order":4,"of":17,"metrics":{"Accuray":"93.62"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-coloninst-v1-seen","task":"Image Classification","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.0 (w/o LoRA, w/o extra data)","rank_in_archive_order":5,"of":17,"metrics":{"Accuray":"93.52"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-coloninst-v1-seen","task":"Image Classification","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.5 (w/ LoRA, w/ extra data)","rank_in_archive_order":17,"of":17,"metrics":{"Accuray":"87.22"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-coloninst-v1-unseen","task":"Image Classification","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.5\n  (w/ LoRA, w/o extra data)","rank_in_archive_order":5,"of":17,"metrics":{"Accuray":"79.24"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-coloninst-v1-unseen","task":"Image Classification","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.0\n  (w/o LoRA, w/o extra data)","rank_in_archive_order":10,"of":17,"metrics":{"Accuray":"78.04"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-coloninst-v1-unseen","task":"Image Classification","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.0\n  (w/o LoRA, w/ extra data)","rank_in_archive_order":12,"of":17,"metrics":{"Accuray":"77.38"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-coloninst-v1-unseen","task":"Image Classification","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.5\n  (w/ LoRA, w/ extra data)","rank_in_archive_order":16,"of":17,"metrics":{"Accuray":"66.51"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst","task":"Referring expression generation","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.5\n  (w/ LoRA, w/o extra data)","rank_in_archive_order":3,"of":17,"metrics":{"Accuray":"99.3"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst","task":"Referring expression generation","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.0\n  (w/o LoRA, w/o extra data)","rank_in_archive_order":9,"of":17,"metrics":{"Accuray":"97.74"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst","task":"Referring expression generation","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.0\n  (w/o LoRA, w/ extra data)","rank_in_archive_order":10,"of":17,"metrics":{"Accuray":"97.35"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst","task":"Referring expression generation","dataset":"ColonINST-v1 (Seen)","model":"LLaVA-Med-v1.5\n  (w/ LoRA, w/ extra data)","rank_in_archive_order":14,"of":17,"metrics":{"Accuray":"90.4"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst-1","task":"Referring expression generation","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.0\n  (w/o LoRA, w/ extra data)","rank_in_archive_order":3,"of":17,"metrics":{"Accuray":"75.25"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst-1","task":"Referring expression generation","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.0\n  (w/o LoRA, w/o extra data)","rank_in_archive_order":5,"of":17,"metrics":{"Accuray":"75.07"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst-1","task":"Referring expression generation","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.5\n  (w/ LoRA, w/o extra data)","rank_in_archive_order":8,"of":17,"metrics":{"Accuray":"73.05"},"uses_additional_data":false},{"leaderboard":"/sota/referring-expression-generation-on-coloninst-1","task":"Referring expression generation","dataset":"ColonINST-v1 (Unseen)","model":"LLaVA-Med-v1.5\n  (w/ LoRA, w/ extra data)","rank_in_archive_order":13,"of":17,"metrics":{"Accuray":"70.00"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2306.00890","atlas_url":"https://app.syntology.ai/?focus=2306.00890","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}