{"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/llama-adapter-v2-parameter-efficient-visual","title":"LLaMA-Adapter V2: Parameter-Efficient Visual Instruction Model","arxiv_id":"2304.15010","date":"2023-04-28","proceeding":null,"authors":["Peng Gao","Jiaming Han","Renrui Zhang","Ziyi Lin","Shijie Geng","Aojun Zhou","Wei zhang","Pan Lu","Conghui He","Xiangyu Yue","Hongsheng Li","Yu Qiao"],"abstract":"How to efficiently transform large language models (LLMs) into instruction followers is recently a popular research direction, while training LLM for multi-modal reasoning remains less explored. Although the recent LLaMA-Adapter demonstrates the potential to handle visual inputs with LLMs, it still cannot generalize well to open-ended visual instructions and lags behind GPT-4. In this paper, we present LLaMA-Adapter V2, a parameter-efficient visual instruction model. Specifically, we first augment LLaMA-Adapter by unlocking more learnable parameters (e.g., norm, bias and scale), which distribute the instruction-following ability across the entire LLaMA model besides adapters. Secondly, we propose an early fusion strategy to feed visual tokens only into the early LLM layers, contributing to better visual knowledge incorporation. Thirdly, a joint training paradigm of image-text pairs and instruction-following data is introduced by optimizing disjoint groups of learnable parameters. This strategy effectively alleviates the interference between the two tasks of image-text alignment and instruction following and achieves strong multi-modal reasoning with only a small-scale image-text and instruction dataset. During inference, we incorporate additional expert models (e.g. captioning/OCR systems) into LLaMA-Adapter to further enhance its image understanding capability without incurring training costs. Compared to the original LLaMA-Adapter, our LLaMA-Adapter V2 can perform open-ended multi-modal instructions by merely introducing 14M parameters over LLaMA. The newly designed framework also exhibits stronger language-only instruction-following capabilities and even excels in chat interactions. Our code and models are available at https://github.com/ZrrSkywalker/LLaMA-Adapter.","url_abs":"https://arxiv.org/abs/2304.15010v1","url_pdf":"https://arxiv.org/pdf/2304.15010v1.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":"llama-adapter-v2-parameter-efficient-visual","repo_url":"https://github.com/zrrskywalker/llama-adapter","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"llama-adapter-v2-parameter-efficient-visual","repo_url":"https://github.com/opengvlab/llama-adapter","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"GPL-3.0"}},{"paper_slug":"llama-adapter-v2-parameter-efficient-visual","repo_url":"https://github.com/Mind23-2/MindCode-140","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":{"status":"ok"}}],"tasks":[{"task_slug":"instruction-following","task_name":"Instruction Following"},{"task_slug":"optical-character-recognition","task_name":"Optical Character Recognition (OCR)"},{"task_slug":"video-question-answering","task_name":"Video Question Answering"},{"task_slug":"video-based-generative-performance","task_name":"Video-based Generative Performance Benchmarking"},{"task_slug":"video-based-generative-performance-5","task_name":"Video-based Generative Performance Benchmarking (Consistency)"},{"task_slug":"video-based-generative-performance-3","task_name":"Video-based Generative Performance Benchmarking (Contextual Understanding)"},{"task_slug":"video-based-generative-performance-1","task_name":"Video-based Generative Performance Benchmarking (Correctness of Information)"},{"task_slug":"video-based-generative-performance-2","task_name":"Video-based Generative Performance Benchmarking (Detail Orientation))"},{"task_slug":"video-based-generative-performance-4","task_name":"Video-based Generative Performance Benchmarking (Temporal Understanding)"},{"task_slug":"visual-question-answering-1","task_name":"Visual Question Answering"},{"task_slug":"visual-question-answering","task_name":"Visual Question Answering (VQA)"},{"task_slug":"zeroshot-video-question-answer","task_name":"Zero-Shot Video Question Answer"},{"task_slug":"model","task_name":"model"}],"methods":[{"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":"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/video-question-answering-on-activitynet-qa","task":"Video Question Answering","dataset":"ActivityNet-QA","model":"LLaMA Adapter V2","rank_in_archive_order":30,"of":36,"metrics":{"Accuracy":"34.2","Confidence score":"2.7"},"uses_additional_data":false},{"leaderboard":"/sota/video-based-generative-performance","task":"Video-based Generative Performance Benchmarking","dataset":"VideoInstruct","model":"LLaMA Adapter","rank_in_archive_order":22,"of":23,"metrics":{"Consistency":"2.15","Contextual Understanding":"2.30","Correctness of Information":"2.03","Detail Orientation":"2.32","Temporal Understanding":"1.98","mean":"2.16"},"uses_additional_data":false},{"leaderboard":"/sota/video-based-generative-performance-2","task":"Video-based Generative Performance Benchmarking (Consistency)","dataset":"VideoInstruct","model":"LLaMA Adapter","rank_in_archive_order":17,"of":18,"metrics":{"gpt-score":"2.15"},"uses_additional_data":false},{"leaderboard":"/sota/video-based-generative-performance-3","task":"Video-based Generative Performance Benchmarking (Contextual Understanding)","dataset":"VideoInstruct","model":"LLaMA Adapter","rank_in_archive_order":17,"of":18,"metrics":{"gpt-score":"2.30"},"uses_additional_data":false},{"leaderboard":"/sota/video-based-generative-performance-1","task":"Video-based Generative Performance Benchmarking (Correctness of Information)","dataset":"VideoInstruct","model":"LLaMA Adapter","rank_in_archive_order":17,"of":18,"metrics":{"gpt-score":"2.03"},"uses_additional_data":false},{"leaderboard":"/sota/video-based-generative-performance-4","task":"Video-based Generative Performance Benchmarking (Detail Orientation))","dataset":"VideoInstruct","model":"LLaMA Adapter","rank_in_archive_order":17,"of":18,"metrics":{"gpt-score":"2.32"},"uses_additional_data":false},{"leaderboard":"/sota/video-based-generative-performance-5","task":"Video-based Generative Performance Benchmarking (Temporal Understanding)","dataset":"VideoInstruct","model":"LLaMA Adapter","rank_in_archive_order":16,"of":18,"metrics":{"gpt-score":"1.98"},"uses_additional_data":false},{"leaderboard":"/sota/visual-question-answering-on-mm-vet","task":"Visual Question Answering","dataset":"MM-Vet","model":"LLaMA-Adapter v2-7B","rank_in_archive_order":194,"of":231,"metrics":{"GPT-4 score":"31.4±0.1","Params":"7B"},"uses_additional_data":false},{"leaderboard":"/sota/visual-question-answering-vqa-on-core-mm","task":"Visual Question Answering (VQA)","dataset":"InfiMM-Eval","model":"LLaMA-Adapter V2","rank_in_archive_order":6,"of":14,"metrics":{"Abductive":"46.12","Analogical":"22.08","Deductive":"28.7","Overall score":"30.46","Params":"7B"},"uses_additional_data":false},{"leaderboard":"/sota/zeroshot-video-question-answer-on-activitynet","task":"Zero-Shot Video Question Answer","dataset":"ActivityNet-QA","model":"LLaMA Adapter","rank_in_archive_order":25,"of":28,"metrics":{"Accuracy":"34.2","Confidence Score":"2.7"},"uses_additional_data":false},{"leaderboard":"/sota/zeroshot-video-question-answer-on-msrvtt-qa","task":"Zero-Shot Video Question Answer","dataset":"MSRVTT-QA","model":"LLaMA Adapter-7B","rank_in_archive_order":29,"of":30,"metrics":{"Accuracy":"43.8","Confidence Score":"2.7"},"uses_additional_data":false},{"leaderboard":"/sota/zeroshot-video-question-answer-on-msvd-qa","task":"Zero-Shot Video Question Answer","dataset":"MSVD-QA","model":"LLaMA Adapter-7B","rank_in_archive_order":26,"of":28,"metrics":{"Accuracy":"54.9","Confidence Score":"3.1"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2304.15010","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2304.15010"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+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":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Mind23-2/MindCode-140","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/opengvlab/llama-adapter","reach":{"status":"ok","spdx":"GPL-3.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/zrrskywalker/llama-adapter","reach":{"status":"ok"}}],"summary":{"unverified":1},"by_repo_kind":{},"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":1,"samples":[{"code_sha256_prefix":"71c517c59227454e","entry":"imagebind_huge","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"71c517c59227454e"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}