{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7QE55K4WABF5FJTSWHR23H5JAB","short_pith_number":"pith:7QE55K4W","schema_version":"1.0","canonical_sha256":"fc09deab96004bd2a672b1e3ad9fa9005644c7c9d92435ed6efa8bf28d92b301","source":{"kind":"arxiv","id":"2608.04548","version":1},"attestation_state":"computed","paper":{"title":"A Model Merging Approach for Continual MLLM Unlearning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haichang Gao, Haoxuan Ji, Linlin Zhang, Xianmin Ye, Yuhang Wang, Zhenxing Niu","submitted_at":"2026-08-05T07:37:23Z","abstract_excerpt":"Multimodal large language model (MLLM) unlearning methods have been proposed to remove private, sensitive, or proprietary information from well-trained models. However, most existing MLLM unlearning methods are designed for one-shot requests and fail to adequately address continual scenarios, as repeatedly applying one-shot operations leads to cumulative utility degradation, unlearning rebound, and retention drift. We introduce Merging for Continual Unlearning (MCU), an approach that dynamically merges multiple one-shot unlearning adapters into a unified adapter upon receiving each new unlearn"},"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":"2608.04548","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-08-05T07:37:23Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8dffc766cb72d7a4b1da22ce09ac5b62cf19af5c4a2575c579695d9f0b8772d5","abstract_canon_sha256":"016924306b4f0c0917d633ae31c0b75ec1d5d4a98c0ea954273408ea22547576"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-06T01:35:49.005533Z","signature_b64":"D51eGjusyhAqUPiaDco26WaoL4KVFXscYaafrg6RAFdbBjf+/u4gIRbAVucLI81qEIOmFvDIZjEI54vCpY5dDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fc09deab96004bd2a672b1e3ad9fa9005644c7c9d92435ed6efa8bf28d92b301","last_reissued_at":"2026-08-06T01:35:49.004137Z","signature_status":"signed_v1","first_computed_at":"2026-08-06T01:35:49.004137Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Model Merging Approach for Continual MLLM Unlearning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haichang Gao, Haoxuan Ji, Linlin Zhang, Xianmin Ye, Yuhang Wang, Zhenxing Niu","submitted_at":"2026-08-05T07:37:23Z","abstract_excerpt":"Multimodal large language model (MLLM) unlearning methods have been proposed to remove private, sensitive, or proprietary information from well-trained models. However, most existing MLLM unlearning methods are designed for one-shot requests and fail to adequately address continual scenarios, as repeatedly applying one-shot operations leads to cumulative utility degradation, unlearning rebound, and retention drift. We introduce Merging for Continual Unlearning (MCU), an approach that dynamically merges multiple one-shot unlearning adapters into a unified adapter upon receiving each new unlearn"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.04548","kind":"arxiv","version":1},"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/2608.04548/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":"2608.04548","created_at":"2026-08-06T01:35:49.005935+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.04548v1","created_at":"2026-08-06T01:35:49.005935+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.04548","created_at":"2026-08-06T01:35:49.005935+00:00"},{"alias_kind":"pith_short_12","alias_value":"7QE55K4WABF5","created_at":"2026-08-06T01:35:49.005935+00:00"},{"alias_kind":"pith_short_16","alias_value":"7QE55K4WABF5FJTS","created_at":"2026-08-06T01:35:49.005935+00:00"},{"alias_kind":"pith_short_8","alias_value":"7QE55K4W","created_at":"2026-08-06T01:35:49.005935+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB","json":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB.json","graph_json":"https://pith.science/api/pith-number/7QE55K4WABF5FJTSWHR23H5JAB/graph.json","events_json":"https://pith.science/api/pith-number/7QE55K4WABF5FJTSWHR23H5JAB/events.json","paper":"https://pith.science/paper/7QE55K4W"},"agent_actions":{"view_html":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB","download_json":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB.json","view_paper":"https://pith.science/paper/7QE55K4W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.04548&json=true","fetch_graph":"https://pith.science/api/pith-number/7QE55K4WABF5FJTSWHR23H5JAB/graph.json","fetch_events":"https://pith.science/api/pith-number/7QE55K4WABF5FJTSWHR23H5JAB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB/action/storage_attestation","attest_author":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB/action/author_attestation","sign_citation":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB/action/citation_signature","submit_replication":"https://pith.science/pith/7QE55K4WABF5FJTSWHR23H5JAB/action/replication_record"}},"created_at":"2026-08-06T01:35:49.005935+00:00","updated_at":"2026-08-06T01:35:49.005935+00:00"}