{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IX3TV7V5TWQP2CH7AZYHGRNAEN","short_pith_number":"pith:IX3TV7V5","schema_version":"1.0","canonical_sha256":"45f73afebd9da0fd08ff06707345a023710f14cd15ec90c2a2477182371a030b","source":{"kind":"arxiv","id":"2607.18615","version":1},"attestation_state":"computed","paper":{"title":"Stochastic Meta-Unlearning: Bridging Language Backbone and Multimodal Unlearning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Gaowen Liu, Jinhao Duan, Sijia Liu, Tianlong Chen, Zijie Liu","submitted_at":"2026-07-21T01:35:33Z","abstract_excerpt":"Machine unlearning for vision-language models (VLMs) remains underexplored. Unlike language models, VLMs combine a language backbone with visual components, which makes unlearning more complex. There is a surprising phenomenon when moving from single-modality unlearning to VLM unlearning: a target forgotten by the standalone language backbone can still be recovered when image information is given to the full VLM. This shows that text-only feedback is not enough for reliable VLM unlearning. Motivated by this observation, we propose Stochastic Meta-Unlearning (SMU), a bilevel framework that uses"},"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":"2607.18615","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-21T01:35:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4fb0e8c0f9039780418304eb8fac22d21531657c42e51ebb8cfb84f7f9431184","abstract_canon_sha256":"e08c315c84d786dc7388cbca0983c3ecb576ed1d69d759ab5b0f685fa8420896"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-22T00:22:56.883124Z","signature_b64":"QViz37NQkN55OYMc5cqzHlGbq8j7YVW0DoML3jSjkpXxDp9rAvS+hbM/1byiisUTeG+GS2nWP4oGm9Zoy2F9Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45f73afebd9da0fd08ff06707345a023710f14cd15ec90c2a2477182371a030b","last_reissued_at":"2026-07-22T00:22:56.882295Z","signature_status":"signed_v1","first_computed_at":"2026-07-22T00:22:56.882295Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stochastic Meta-Unlearning: Bridging Language Backbone and Multimodal Unlearning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Gaowen Liu, Jinhao Duan, Sijia Liu, Tianlong Chen, Zijie Liu","submitted_at":"2026-07-21T01:35:33Z","abstract_excerpt":"Machine unlearning for vision-language models (VLMs) remains underexplored. Unlike language models, VLMs combine a language backbone with visual components, which makes unlearning more complex. There is a surprising phenomenon when moving from single-modality unlearning to VLM unlearning: a target forgotten by the standalone language backbone can still be recovered when image information is given to the full VLM. This shows that text-only feedback is not enough for reliable VLM unlearning. Motivated by this observation, we propose Stochastic Meta-Unlearning (SMU), a bilevel framework that uses"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.18615","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/2607.18615/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":"2607.18615","created_at":"2026-07-22T00:22:56.882725+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.18615v1","created_at":"2026-07-22T00:22:56.882725+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.18615","created_at":"2026-07-22T00:22:56.882725+00:00"},{"alias_kind":"pith_short_12","alias_value":"IX3TV7V5TWQP","created_at":"2026-07-22T00:22:56.882725+00:00"},{"alias_kind":"pith_short_16","alias_value":"IX3TV7V5TWQP2CH7","created_at":"2026-07-22T00:22:56.882725+00:00"},{"alias_kind":"pith_short_8","alias_value":"IX3TV7V5","created_at":"2026-07-22T00:22:56.882725+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/IX3TV7V5TWQP2CH7AZYHGRNAEN","json":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN.json","graph_json":"https://pith.science/api/pith-number/IX3TV7V5TWQP2CH7AZYHGRNAEN/graph.json","events_json":"https://pith.science/api/pith-number/IX3TV7V5TWQP2CH7AZYHGRNAEN/events.json","paper":"https://pith.science/paper/IX3TV7V5"},"agent_actions":{"view_html":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN","download_json":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN.json","view_paper":"https://pith.science/paper/IX3TV7V5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.18615&json=true","fetch_graph":"https://pith.science/api/pith-number/IX3TV7V5TWQP2CH7AZYHGRNAEN/graph.json","fetch_events":"https://pith.science/api/pith-number/IX3TV7V5TWQP2CH7AZYHGRNAEN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN/action/storage_attestation","attest_author":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN/action/author_attestation","sign_citation":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN/action/citation_signature","submit_replication":"https://pith.science/pith/IX3TV7V5TWQP2CH7AZYHGRNAEN/action/replication_record"}},"created_at":"2026-07-22T00:22:56.882725+00:00","updated_at":"2026-07-22T00:22:56.882725+00:00"}