{"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/vlis-unimodal-language-models-guide","title":"VLIS: Unimodal Language Models Guide Multimodal Language Generation","arxiv_id":"2310.09767","date":"2023-10-15","proceeding":null,"authors":["Jiwan Chung","Youngjae Yu"],"abstract":"Multimodal language generation, which leverages the synergy of language and vision, is a rapidly expanding field. However, existing vision-language models face challenges in tasks that require complex linguistic understanding. To address this issue, we introduce Visual-Language models as Importance Sampling weights (VLIS), a novel framework that combines the visual conditioning capability of vision-language models with the language understanding of unimodal text-only language models without further training. It extracts pointwise mutual information of each image and text from a visual-language model and uses the value as an importance sampling weight to adjust the token likelihood from a text-only model. VLIS improves vision-language models on diverse tasks, including commonsense understanding (WHOOPS, OK-VQA, and ScienceQA) and complex text generation (Concadia, Image Paragraph Captioning, and ROCStories). Our results suggest that VLIS represents a promising new direction for multimodal language generation.","url_abs":"https://arxiv.org/abs/2310.09767v2","url_pdf":"https://arxiv.org/pdf/2310.09767v2.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":"vlis-unimodal-language-models-guide","repo_url":"https://github.com/jiwanchung/vlis","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"caption-generation","task_name":"Caption Generation"},{"task_slug":"explanation-generation","task_name":"Explanation Generation"},{"task_slug":"image-paragraph-captioning","task_name":"Image Paragraph Captioning"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"text-generation","task_name":"Text Generation"},{"task_slug":"visual-question-answering","task_name":"Visual Question Answering (VQA)"},{"task_slug":"zero-shot-image-paragraph-captioning","task_name":"Zero-Shot Image Paragraph Captioning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/explanation-generation-on-whoops","task":"Explanation Generation","dataset":"WHOOPS!","model":"VLIS (Lynx)","rank_in_archive_order":6,"of":7,"metrics":{"Accuracy":"80"},"uses_additional_data":false},{"leaderboard":"/sota/explanation-generation-on-whoops","task":"Explanation Generation","dataset":"WHOOPS!","model":"VLIS (LLaVA)","rank_in_archive_order":7,"of":7,"metrics":{"Accuracy":"73"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2310.09767","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}