{"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/towards-modular-llms-by-building-and-reusing","title":"Towards Modular LLMs by Building and Reusing a Library of LoRAs","arxiv_id":"2405.11157","date":"2024-05-18","proceeding":null,"authors":["Oleksiy Ostapenko","Zhan Su","Edoardo Maria Ponti","Laurent Charlin","Nicolas Le Roux","Matheus Pereira","Lucas Caccia","Alessandro Sordoni"],"abstract":"The growing number of parameter-efficient adaptations of a base large language model (LLM) calls for studying whether we can reuse such trained adapters to improve performance for new tasks. We study how to best build a library of adapters given multi-task data and devise techniques for both zero-shot and supervised task generalization through routing in such library. We benchmark existing approaches to build this library and introduce model-based clustering, MBC, a method that groups tasks based on the similarity of their adapter parameters, indirectly optimizing for transfer across the multi-task dataset. To re-use the library, we present a novel zero-shot routing mechanism, Arrow, which enables dynamic selection of the most relevant adapters for new inputs without the need for retraining. We experiment with several LLMs, such as Phi-2 and Mistral, on a wide array of held-out tasks, verifying that MBC-based adapters and Arrow routing lead to superior generalization to new tasks. We make steps towards creating modular, adaptable LLMs that can match or outperform traditional joint training.","url_abs":"https://arxiv.org/abs/2405.11157v1","url_pdf":"https://arxiv.org/pdf/2405.11157v1.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":"towards-modular-llms-by-building-and-reusing","repo_url":"https://github.com/microsoft/mttl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"large-language-model","task_name":"Large Language Model"}],"methods":[{"method_slug":"adapter","method_name":"Adapter"},{"method_slug":"base","method_name":"BASE"},{"method_slug":null,"method_name":"Library"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2405.11157","atlas_url":"https://app.syntology.ai/?focus=2405.11157","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2405.11157"}},"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/microsoft/mttl","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":13,"unverified":1},"by_repo_kind":{"listed":{"samples":14,"ran":13,"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":"9607e74609d06a5c","entry":"augment_few_shot_task","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mt_seq_to_seq_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mt_seq_to_seq_module.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"9607e74609d06a5c"}},{"code_sha256_prefix":"91c46d2158999132","entry":"augment_prompts","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mmlu_data_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mmlu_data_module.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"91c46d2158999132"}},{"code_sha256_prefix":"90b13d358f062991","entry":"completion_template","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/humaneval_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/humaneval_module.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"90b13d358f062991"}},{"code_sha256_prefix":"3d1190e58b358dda","entry":"completion_template","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mbpp_datamodule.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mbpp_datamodule.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"3d1190e58b358dda"}},{"code_sha256_prefix":"0a70817429b93f25","entry":"create_config_class_from_args","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/arguments.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/arguments.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0a70817429b93f25"}},{"code_sha256_prefix":"801642274b17ca93","entry":"detect_indentation","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mbpp_datamodule.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mbpp_datamodule.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"801642274b17ca93"}},{"code_sha256_prefix":"ef9fcdcd4db23c46","entry":"instruct_template","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/humaneval_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/humaneval_module.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"ef9fcdcd4db23c46"}},{"code_sha256_prefix":"d549f96aaac26455","entry":"instruct_template","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mbpp_datamodule.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mbpp_datamodule.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"d549f96aaac26455"}},{"code_sha256_prefix":"ea35b5a896e6de9d","entry":"is_phi2_eval_task","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mt_seq_to_seq_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mt_seq_to_seq_module.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"ea35b5a896e6de9d"}},{"code_sha256_prefix":"8df019af71f4ed70","entry":"maybe_filter_hf_dataset_by_key","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/clip_data_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/clip_data_module.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"8df019af71f4ed70"}},{"code_sha256_prefix":"0a3ee7f35121b4ea","entry":"permute_options","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mmlu_data_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mmlu_data_module.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0a3ee7f35121b4ea"}},{"code_sha256_prefix":"6b3aac9749e83a24","entry":"subsample_dst","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/base.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/base.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"6b3aac9749e83a24"}},{"code_sha256_prefix":"f7e24d6e9c199261","entry":"take_n_examples_per_task","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/base.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/base.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f7e24d6e9c199261"}},{"code_sha256_prefix":"6773eed2108a3c15","entry":"augment_few_shot","repo":"microsoft/mttl","repo_kind":"listed","path":"mttl/datamodule/mt_seq_to_seq_module.py","file_url":"https://github.com/microsoft/mttl/blob/HEAD/mttl/datamodule/mt_seq_to_seq_module.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"6773eed2108a3c15"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}