{"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/dataless-knowledge-fusion-by-merging-weights","title":"Dataless Knowledge Fusion by Merging Weights of Language Models","arxiv_id":"2212.09849","date":"2022-12-19","proceeding":null,"authors":["Xisen Jin","Xiang Ren","Daniel Preotiuc-Pietro","Pengxiang Cheng"],"abstract":"Fine-tuning pre-trained language models has become the prevalent paradigm for building downstream NLP models. Oftentimes fine-tuned models are readily available but their training data is not, due to data privacy or intellectual property concerns. This creates a barrier to fusing knowledge across individual models to yield a better single model. In this paper, we study the problem of merging individual models built on different training data sets to obtain a single model that performs well both across all data set domains and can generalize on out-of-domain data. We propose a dataless knowledge fusion method that merges models in their parameter space, guided by weights that minimize prediction differences between the merged model and the individual models. Over a battery of evaluation settings, we show that the proposed method significantly outperforms baselines such as Fisher-weighted averaging or model ensembling. Further, we find that our method is a promising alternative to multi-task learning that can preserve or sometimes improve over the individual models without access to the training data. Finally, model merging is more efficient than training a multi-task model, thus making it applicable to a wider set of scenarios.","url_abs":"https://arxiv.org/abs/2212.09849v5","url_pdf":"https://arxiv.org/pdf/2212.09849v5.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":"dataless-knowledge-fusion-by-merging-weights","repo_url":"https://github.com/bloomberg/dataless-model-merging","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"multi-task-learning","task_name":"Multi-Task Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2212.09849","atlas_url":"https://app.syntology.ai/?focus=2212.09849","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2212.09849"}},"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/bloomberg/dataless-model-merging","reach":null}],"summary":{"ran":4,"ran_draft_wrong":6,"unverified":9},"by_repo_kind":{"official":{"samples":19,"ran":10,"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":"10b9df811ddc6d24","entry":"FedAvgMerger","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"10b9df811ddc6d24"}},{"code_sha256_prefix":"f3635898362e71cb","entry":"GroundMetric","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"f3635898362e71cb"}},{"code_sha256_prefix":"c314ef759f51e47d","entry":"ModelMergerBase","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c314ef759f51e47d"}},{"code_sha256_prefix":"e823833341bb66b2","entry":"TmpLocalModel","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e823833341bb66b2"}},{"code_sha256_prefix":"9512ea87c384d970","entry":"compute_bce_loss","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9512ea87c384d970"}},{"code_sha256_prefix":"af8fc9a6b079dd80","entry":"filter_modules_by_regex","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"af8fc9a6b079dd80"}},{"code_sha256_prefix":"542bdbc6b44e26fe","entry":"filter_params_to_merge","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"542bdbc6b44e26fe"}},{"code_sha256_prefix":"b7a34bad497441a4","entry":"get_component_configs","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"b7a34bad497441a4"}},{"code_sha256_prefix":"d150cc5adcf9981b","entry":"get_reweight_factor","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"d150cc5adcf9981b"}},{"code_sha256_prefix":"253168af406c0497","entry":"get_submodule","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"253168af406c0497"}},{"code_sha256_prefix":"9f88183c29de2263","entry":"DebertaV2ForMultiLabelClassification","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9f88183c29de2263"}},{"code_sha256_prefix":"cb4e87b0878653c2","entry":"DistilBERTForMultiLabelClassification","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"cb4e87b0878653c2"}},{"code_sha256_prefix":"dec050be98e95d6a","entry":"GPT2ForMultiLabelClassification","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"dec050be98e95d6a"}},{"code_sha256_prefix":"9410ed9ff867ec7d","entry":"MTLModel","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9410ed9ff867ec7d"}},{"code_sha256_prefix":"80ffc739f078f603","entry":"MultiDomainModel","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"80ffc739f078f603"}},{"code_sha256_prefix":"cc27cfd051c590d3","entry":"OptimalTransportMerger","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"cc27cfd051c590d3"}},{"code_sha256_prefix":"7ff12c9ca05e8578","entry":"RoBERTaForMultiLabelClassification","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7ff12c9ca05e8578"}},{"code_sha256_prefix":"3b2e0a4e703d6201","entry":"create_model","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"3b2e0a4e703d6201"}},{"code_sha256_prefix":"4b49e7003c26de74","entry":"create_stl_model","repo":"bloomberg/dataless-model-merging","repo_kind":"official","path":"src/model_merge/ot_merger.py","file_url":"https://github.com/bloomberg/dataless-model-merging/blob/HEAD/src/model_merge/ot_merger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"4b49e7003c26de74"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}