{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GJMK4BY6MWD3AQ3IJQBYH7LHIL","short_pith_number":"pith:GJMK4BY6","schema_version":"1.0","canonical_sha256":"3258ae071e6587b043684c0383fd6742c3bfa98eef24dd7299b6692eb428f531","source":{"kind":"arxiv","id":"2502.12420","version":2},"attestation_state":"computed","paper":{"title":"Sens-Merging: Sensitivity-Guided Parameter Balancing for Merging Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bowei He, Han Wu, Linqi Song, Mingxuan Yuan, Shuqi Liu, Xiongwei Han","submitted_at":"2025-02-18T01:41:13Z","abstract_excerpt":"Recent advances in large language models have led to numerous task-specialized fine-tuned variants, creating a need for efficient model merging techniques that preserve specialized capabilities while avoiding costly retraining. While existing task vector-based merging methods show promise, they typically apply uniform coefficients across all parameters, overlooking varying parameter importance both within and across tasks. We present Sens-Merging, a sensitivity-guided coefficient adjustment method that enhances existing model merging techniques by operating at both task-specific and cross-task"},"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":"2502.12420","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-18T01:41:13Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"0c29ff663390c05ac7cc4b54fd2f40abee6e1f33c06d288298fa6ccadabc62e7","abstract_canon_sha256":"3a3be4b3bee06bbee8306efa620ae26148e4956b08bbf5b81e6c0a84d0f75fac"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:16:42.568856Z","signature_b64":"HaM7Ss22RH+rVhULBTfpsc9YZ+PliaAcK2KUforbZ5Xq7fI/9vxYINbufVLoYuoeMmlLYO68CZLijySK/pIBCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3258ae071e6587b043684c0383fd6742c3bfa98eef24dd7299b6692eb428f531","last_reissued_at":"2026-07-05T10:16:42.568278Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:16:42.568278Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sens-Merging: Sensitivity-Guided Parameter Balancing for Merging Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bowei He, Han Wu, Linqi Song, Mingxuan Yuan, Shuqi Liu, Xiongwei Han","submitted_at":"2025-02-18T01:41:13Z","abstract_excerpt":"Recent advances in large language models have led to numerous task-specialized fine-tuned variants, creating a need for efficient model merging techniques that preserve specialized capabilities while avoiding costly retraining. While existing task vector-based merging methods show promise, they typically apply uniform coefficients across all parameters, overlooking varying parameter importance both within and across tasks. We present Sens-Merging, a sensitivity-guided coefficient adjustment method that enhances existing model merging techniques by operating at both task-specific and cross-task"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.12420","kind":"arxiv","version":2},"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/2502.12420/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":"2502.12420","created_at":"2026-07-05T10:16:42.568346+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.12420v2","created_at":"2026-07-05T10:16:42.568346+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.12420","created_at":"2026-07-05T10:16:42.568346+00:00"},{"alias_kind":"pith_short_12","alias_value":"GJMK4BY6MWD3","created_at":"2026-07-05T10:16:42.568346+00:00"},{"alias_kind":"pith_short_16","alias_value":"GJMK4BY6MWD3AQ3I","created_at":"2026-07-05T10:16:42.568346+00:00"},{"alias_kind":"pith_short_8","alias_value":"GJMK4BY6","created_at":"2026-07-05T10:16:42.568346+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.29489","citing_title":"Access Sets Matter: Budgeting Expert Reads for Scalable Weight-Space Model Merging","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL","json":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL.json","graph_json":"https://pith.science/api/pith-number/GJMK4BY6MWD3AQ3IJQBYH7LHIL/graph.json","events_json":"https://pith.science/api/pith-number/GJMK4BY6MWD3AQ3IJQBYH7LHIL/events.json","paper":"https://pith.science/paper/GJMK4BY6"},"agent_actions":{"view_html":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL","download_json":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL.json","view_paper":"https://pith.science/paper/GJMK4BY6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.12420&json=true","fetch_graph":"https://pith.science/api/pith-number/GJMK4BY6MWD3AQ3IJQBYH7LHIL/graph.json","fetch_events":"https://pith.science/api/pith-number/GJMK4BY6MWD3AQ3IJQBYH7LHIL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL/action/storage_attestation","attest_author":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL/action/author_attestation","sign_citation":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL/action/citation_signature","submit_replication":"https://pith.science/pith/GJMK4BY6MWD3AQ3IJQBYH7LHIL/action/replication_record"}},"created_at":"2026-07-05T10:16:42.568346+00:00","updated_at":"2026-07-05T10:16:42.568346+00:00"}