{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ENBESPZ4EMWAFUDTY45RWOHJLA","short_pith_number":"pith:ENBESPZ4","schema_version":"1.0","canonical_sha256":"2342493f3c232c02d073c73b1b38e95802db1419f7d30a24301f03d760b67f3d","source":{"kind":"arxiv","id":"2302.07027","version":3},"attestation_state":"computed","paper":{"title":"AdapterSoup: Weight Averaging to Improve Generalization of Pretrained Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alexander Fraser, Alexandra Chronopoulou, Jesse Dodge, Matthew E. Peters","submitted_at":"2023-02-14T13:09:23Z","abstract_excerpt":"Pretrained language models (PLMs) are trained on massive corpora, but often need to specialize to specific domains. A parameter-efficient adaptation method suggests training an adapter for each domain on the task of language modeling. This leads to good in-domain scores but can be impractical for domain- or resource-restricted settings. A solution is to use a related-domain adapter for the novel domain at test time. In this paper, we introduce AdapterSoup, an approach that performs weight-space averaging of adapters trained on different domains. Our approach is embarrassingly parallel: first, "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2302.07027","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-14T13:09:23Z","cross_cats_sorted":[],"title_canon_sha256":"8536b8b4e9974fc403e8873b9cc5de2f9e42fc813e0ef2649919ddf9b6343a41","abstract_canon_sha256":"9cbb1bebd48292aed3b249a9e49838bbfc1ab22537bb55e8500ab491dfa1368b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:55:19.144421Z","signature_b64":"QHAP4Lde1eISyIHCKkW/0yT6hldbzBM55q1xhbf7mllRJynSr5q5dCN7uwNwfNsAljfOYBnXc+xIEF5CFzrCAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2342493f3c232c02d073c73b1b38e95802db1419f7d30a24301f03d760b67f3d","last_reissued_at":"2026-07-05T05:55:19.143982Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:55:19.143982Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AdapterSoup: Weight Averaging to Improve Generalization of Pretrained Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alexander Fraser, Alexandra Chronopoulou, Jesse Dodge, Matthew E. Peters","submitted_at":"2023-02-14T13:09:23Z","abstract_excerpt":"Pretrained language models (PLMs) are trained on massive corpora, but often need to specialize to specific domains. A parameter-efficient adaptation method suggests training an adapter for each domain on the task of language modeling. This leads to good in-domain scores but can be impractical for domain- or resource-restricted settings. A solution is to use a related-domain adapter for the novel domain at test time. In this paper, we introduce AdapterSoup, an approach that performs weight-space averaging of adapters trained on different domains. Our approach is embarrassingly parallel: first, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.07027","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2302.07027/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2302.07027","created_at":"2026-07-05T05:55:19.144042+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.07027v3","created_at":"2026-07-05T05:55:19.144042+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.07027","created_at":"2026-07-05T05:55:19.144042+00:00"},{"alias_kind":"pith_short_12","alias_value":"ENBESPZ4EMWA","created_at":"2026-07-05T05:55:19.144042+00:00"},{"alias_kind":"pith_short_16","alias_value":"ENBESPZ4EMWAFUDT","created_at":"2026-07-05T05:55:19.144042+00:00"},{"alias_kind":"pith_short_8","alias_value":"ENBESPZ4","created_at":"2026-07-05T05:55:19.144042+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2403.14608","citing_title":"Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey","ref_index":38,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA","json":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA.json","graph_json":"https://pith.science/api/pith-number/ENBESPZ4EMWAFUDTY45RWOHJLA/graph.json","events_json":"https://pith.science/api/pith-number/ENBESPZ4EMWAFUDTY45RWOHJLA/events.json","paper":"https://pith.science/paper/ENBESPZ4"},"agent_actions":{"view_html":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA","download_json":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA.json","view_paper":"https://pith.science/paper/ENBESPZ4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.07027&json=true","fetch_graph":"https://pith.science/api/pith-number/ENBESPZ4EMWAFUDTY45RWOHJLA/graph.json","fetch_events":"https://pith.science/api/pith-number/ENBESPZ4EMWAFUDTY45RWOHJLA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA/action/storage_attestation","attest_author":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA/action/author_attestation","sign_citation":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA/action/citation_signature","submit_replication":"https://pith.science/pith/ENBESPZ4EMWAFUDTY45RWOHJLA/action/replication_record"}},"created_at":"2026-07-05T05:55:19.144042+00:00","updated_at":"2026-07-05T05:55:19.144042+00:00"}