{"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/asymmetry-in-low-rank-adapters-of-foundation","title":"Asymmetry in Low-Rank Adapters of Foundation Models","arxiv_id":"2402.16842","date":"2024-02-26","proceeding":null,"authors":["Jiacheng Zhu","Kristjan Greenewald","Kimia Nadjahi","Haitz Sáez de Ocáriz Borde","Rickard Brüel Gabrielsson","Leshem Choshen","Marzyeh Ghassemi","Mikhail Yurochkin","Justin Solomon"],"abstract":"Parameter-efficient fine-tuning optimizes large, pre-trained foundation models by updating a subset of parameters; in this class, Low-Rank Adaptation (LoRA) is particularly effective. Inspired by an effort to investigate the different roles of LoRA matrices during fine-tuning, this paper characterizes and leverages unexpected asymmetry in the importance of low-rank adapter matrices. Specifically, when updating the parameter matrices of a neural network by adding a product $BA$, we observe that the $B$ and $A$ matrices have distinct functions: $A$ extracts features from the input, while $B$ uses these features to create the desired output. Based on this observation, we demonstrate that fine-tuning $B$ is inherently more effective than fine-tuning $A$, and that a random untrained $A$ should perform nearly as well as a fine-tuned one. Using an information-theoretic lens, we also bound the generalization of low-rank adapters, showing that the parameter savings of exclusively training $B$ improves the bound. We support our conclusions with experiments on RoBERTa, BART-Large, LLaMA-2, and ViTs.","url_abs":"https://arxiv.org/abs/2402.16842v2","url_pdf":"https://arxiv.org/pdf/2402.16842v2.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":"asymmetry-in-low-rank-adapters-of-foundation","repo_url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"parameter-efficient-fine-tuning","task_name":"parameter-efficient fine-tuning"}],"methods":[{"method_slug":"adam","method_name":"Adam"},{"method_slug":"adapter","method_name":"Adapter"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"attention-dropout","method_name":"Attention Dropout"},{"method_slug":"bert","method_name":"BERT"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"linear-warmup-with-linear-decay","method_name":"Linear Warmup With Linear Decay"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"roberta","method_name":"RoBERTa"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"weight-decay","method_name":"Weight Decay"},{"method_slug":"wordpiece","method_name":"WordPiece"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2402.16842","atlas_url":"https://app.syntology.ai/?focus=2402.16842","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2402.16842"}},"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":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA","reach":{"status":"ok"}}],"summary":{"ran":5,"unverified":1},"by_repo_kind":{"official":{"samples":6,"ran":5,"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":6,"samples":[{"code_sha256_prefix":"875f9e0e658b6c6d","entry":"STRATEGY_TO_RULE_func","repo":"Jiacheng-Zhu-AIML/AsymmetryLoRA","repo_kind":"official","path":"GPT_experiments/local_lora_configs.py","file_url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA/blob/HEAD/GPT_experiments/local_lora_configs.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"875f9e0e658b6c6d"}},{"code_sha256_prefix":"96556814f1bac2ea","entry":"get_peft_model_state_dict_local","repo":"Jiacheng-Zhu-AIML/AsymmetryLoRA","repo_kind":"official","path":"LoRASYM_peft/local_peft_model_all.py","file_url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA/blob/HEAD/LoRASYM_peft/local_peft_model_all.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"96556814f1bac2ea"}},{"code_sha256_prefix":"59f97c4aab5b76e8","entry":"match","repo":"Jiacheng-Zhu-AIML/AsymmetryLoRA","repo_kind":"official","path":"GPT_experiments/utils.py","file_url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA/blob/HEAD/GPT_experiments/utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"59f97c4aab5b76e8"}},{"code_sha256_prefix":"020b699ab74fb7e1","entry":"process_mcq","repo":"Jiacheng-Zhu-AIML/AsymmetryLoRA","repo_kind":"official","path":"GPT_experiments/utils.py","file_url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA/blob/HEAD/GPT_experiments/utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"020b699ab74fb7e1"}},{"code_sha256_prefix":"d8ca1158aeb288e1","entry":"strategy_to_rule_dict","repo":"Jiacheng-Zhu-AIML/AsymmetryLoRA","repo_kind":"official","path":"GPT_experiments/local_lora_configs.py","file_url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA/blob/HEAD/GPT_experiments/local_lora_configs.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"d8ca1158aeb288e1"}},{"code_sha256_prefix":"2200fea63d2ae407","entry":"measure_accuracy_by_category","repo":"Jiacheng-Zhu-AIML/AsymmetryLoRA","repo_kind":"official","path":"GPT_experiments/utils.py","file_url":"https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA/blob/HEAD/GPT_experiments/utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"2200fea63d2ae407"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}