{"as_of":"2026-08-23T10:40:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:22d800c41ae8c0a0d0daccc93bb5af6e747b3cc9753db7b4e4a6e82e70a36a5b","coverage":[{"denominator":70,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":70,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-04T11:27:23.020889Z","state":"measured"},{"denominator":72,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":72,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-23T06:30:58.430688+00:00","state":"measured"},{"denominator":2,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":2,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-15T14:17:57.173259Z","state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"arxiv_reference","source_observed_at":"2026-05-19T21:52:48.286928Z","state":"measured"}],"external_citation_measurements":[],"inbound":[{"citation":{"cited_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"cited_work":{"arxiv_id":"2510.04773","doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":"2510.04773","snapshot_observed_at":"2026-06-25T02:17:31.129069Z","title":"arXiv preprint arXiv:2510.04773 , year=","venue":null,"work_id":"e5ecc2d9-eb2a-498b-8178-5788d50d8b28","year":null},"citing_paper":{"arxiv_id":"2605.16776","last_updated":"2026-05-16T03:15:35Z","snapshot_observed_at":"2026-08-14T10:46:02.829450Z","submitted_at":"2026-05-16T03:15:35Z","title":"Distinguishable Deletion: Unifying Knowledge Erasure and Refusal for Large Language Model Unlearning","version":1},"reference_index":120,"source":"arxiv_source","source_observed_at":"2026-05-19T21:49:07.440832Z"},"links":{"cited_paper":"/paper/2510.04773","citing_paper":"/paper/2605.16776"},"observation_digest":"sha256:becca49b418362f0e5e90c17f7584d42db4a39386cf9420ab9a5272e50385df7","observation_id":"584746ad-b1e1-4758-b4e4-2bbffacf8029","resolution":{"observed_at":"2026-06-25T02:17:31.129069Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-23T06:30:58.430688+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-23T06:30:58.430688+00:00","source":"crossref"},{"observed_at":"2026-08-23T06:30:53.778098+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2510.04773","snapshot_observed_at":"2026-08-15T14:17:57.173259Z","title":"arXiv preprint arXiv:2510.04773 , year=","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.11408","last_updated":"2026-08-11T20:16:20Z","snapshot_observed_at":"2026-08-21T16:01:22.083634Z","submitted_at":"2026-08-11T20:16:20Z","title":"Measure, Don't Optimize: Forecasting Recovery in LLM Unlearning","version":1},"reference_index":191,"source":"arxiv_source","source_observed_at":"2026-08-15T14:17:57.173259Z"},"links":{"cited_paper":"/paper/2510.04773","citing_paper":"/paper/2608.11408"},"observation_digest":"sha256:a611edc8c1935990d9822f2774aea512fb46621770cc7814e205465f39724300","observation_id":"ad72ebad-475a-4e0b-81a9-969dc3ec1968","resolution":{"observed_at":"2026-08-15T14:17:57.173259Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"links":{"evidence":"/evidence","html":"/paper/2510.04773/citation-record","integrity":"/paper/2510.04773/integrity","json":"/paper/2510.04773/citation-record.json","paper":"/paper/2510.04773"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:20.452230Z","title":"Unlearning bias in language models by partitioning gradients","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.452230Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:959ca1d36a98391178aad778d13aed67f987301c076c3322410872978ab0f9dd","observation_id":"76d60a61-f8ad-4845-aa45-3f4f255b4314","resolution":{"observed_at":"2026-08-04T11:27:20.452230Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:20.531377Z","title":"Continual learning and private unlearning","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.531377Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:723de28e489a875e3d369ae72e5e2a5139d47f74577ee8352eae2fbf7e5b388b","observation_id":"a7549dde-221c-4458-ba98-927e18a0d670","resolution":{"observed_at":"2026-08-04T11:27:20.531377Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:20.622372Z","title":"Who’s harry potter? approximate unlearning for llms","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.622372Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:7a8f86436c05bcf371114649a208b1fe82729cdd01557be6c47d5e39ed366aba","observation_id":"b2ff310c-a1d8-4d32-9566-473d2a7306c9","resolution":{"observed_at":"2026-08-04T11:27:20.622372Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2210.01504","last_updated":"2022-12-19T11:52:40Z","snapshot_observed_at":"2026-08-16T16:27:04.380829Z","submitted_at":"2022-10-04T10:18:11Z","title":"Knowledge Unlearning for Mitigating Privacy Risks in Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2210.01504","snapshot_observed_at":"2026-08-04T11:27:20.679549Z","title":"Knowledge unlearning for mitigating privacy risks in language models.arXiv preprint arXiv:2210.01504, 2022","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.679549Z"},"links":{"cited_paper":"/paper/2210.01504","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:465e0d4b937fc84995596861ad1b84cbd4908ce7f3b7bfe3077f0521b8a5c62f","observation_id":"8a903e82-8889-48cc-97a6-3f6827892dcf","resolution":{"observed_at":"2026-08-04T11:27:20.679549Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.20138","last_updated":"2023-12-05T16:14:24Z","snapshot_observed_at":"2026-08-19T21:06:59.277994Z","submitted_at":"2023-10-31T03:09:36Z","title":"DEPN: Detecting and Editing Privacy Neurons in Pretrained Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.20138","snapshot_observed_at":"2026-08-04T11:27:20.743364Z","title":"Depn: Detecting and editing privacy neurons in pretrained language models.arXiv preprint arXiv:2310.20138, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.743364Z"},"links":{"cited_paper":"/paper/2310.20138","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:71febe0803fc8baab6f23b557cc6a67723e149bcedbdc33af9d5d84db66459f4","observation_id":"ab763c55-6373-4ee0-a0ec-7493f909c78e","resolution":{"observed_at":"2026-08-04T11:27:20.743364Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2402.10058","last_updated":"2024-06-05T04:57:09Z","snapshot_observed_at":"2026-08-19T00:25:45.368961Z","submitted_at":"2024-02-15T16:28:34Z","title":"Towards Safer Large Language Models through Machine Unlearning","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2402.10058","snapshot_observed_at":"2026-08-04T11:27:20.798918Z","title":"Towards safer large language models through machine unlearning.arXiv preprint arXiv:2402.10058, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.798918Z"},"links":{"cited_paper":"/paper/2402.10058","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:1f0c382068a8eb85cc53e87102e019d602936917299f22103f4ba34d13bbbb48","observation_id":"d68c6b24-d6c9-4f77-b76d-b49076c2e403","resolution":{"observed_at":"2026-08-04T11:27:20.798918Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:20.870907Z","title":"Identifying and mitigating the security risks of generative ai.Foundations and Trends® in Privacy and Security, 6(1):1–52, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.870907Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:4e71efa4886b84e25706fb6a60a3a16ef5c7d7c27a451ae2de59bdf8d180524f","observation_id":"f5e3ef6c-75e9-43b7-aa67-8bd9060fcbaa","resolution":{"observed_at":"2026-08-04T11:27:20.870907Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:20.943176Z","title":null,"venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:20.943176Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:738b781f0f663dd7df7b463c99755ca85c4a0cb4a4a84876599b4e7bfd7a9786","observation_id":"a8f923d0-7a5c-4a63-8d89-43b3f4c3866f","resolution":{"observed_at":"2026-08-04T11:27:20.943176Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.028350Z","title":"Assembly bill no","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.028350Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:90a483107a560172b4458cdc3cb65ac7bc2b0cf8b4b33a57fcc820ca1e021e47","observation_id":"773521d1-8438-45a1-8842-97cced60dc55","resolution":{"observed_at":"2026-08-04T11:27:21.028350Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.177914Z","title":"Large language model unlearning.Advances in Neural Information Processing Systems, 37:105425–105475, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.177914Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:596e0347c35b671812a9c07416550c2c40750747911c833ba536477c077d5f10","observation_id":"d06458fb-c61e-48ae-b485-bfdd3a2481b8","resolution":{"observed_at":"2026-08-04T11:27:21.177914Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.305220Z","title":"Negative preference optimization: From catastrophic collapse to effective unlearning","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.305220Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:60ac832d6f4b5ae54b91387027da4cb12f029db34833615d11212d5618177c23","observation_id":"67aec5fc-288a-4ef4-bf2e-9a71c26d0c3a","resolution":{"observed_at":"2026-08-04T11:27:21.305220Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.398045Z","title":"Direct preference optimization: Your language model is secretly a reward model","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.398045Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:d0e228a6d9d985a10d235df6293acdc38d86dd1892ea5598d9c9efd4a6ef2204","observation_id":"62129147-0ca8-467b-bb76-2c52c5b57dc8","resolution":{"observed_at":"2026-08-04T11:27:21.398045Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.500262Z","title":"TOFU: A task of fictitious unlearning for LLMs","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.500262Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:e82fff5ce982b62c144dd2350ca520c48fc9dba1488dd1e18ba3fa460356d6e3","observation_id":"c84421ff-e3ad-45d2-82b8-10c0d87783c3","resolution":{"observed_at":"2026-08-04T11:27:21.500262Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.569482Z","title":"Smith, and Chiyuan Zhang","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.569482Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:a44397590a8dc799ee9be735b3ee0415321571e9e64edd52aedbfd8a3b996efe","observation_id":"b275dbd2-761a-43ec-9639-83242c7c3669","resolution":{"observed_at":"2026-08-04T11:27:21.569482Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2209.02299","last_updated":"2024-09-17T11:55:58Z","snapshot_observed_at":"2026-08-18T20:41:01.318983Z","submitted_at":"2022-09-06T08:51:53Z","title":"A Survey of Machine Unlearning","version":6},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2209.02299","snapshot_observed_at":"2026-08-04T11:27:21.659846Z","title":"A survey of machine unlearning.arXiv preprint arXiv:2209.02299, 2022","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.659846Z"},"links":{"cited_paper":"/paper/2209.02299","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:cd91352e11b25dd3d04f7b1805eba7aa0cda0cc34aa1b97556e5ce3b6552f2b5","observation_id":"8ad36c83-7027-4ce1-bd45-4a21c8f3cdf0","resolution":{"observed_at":"2026-08-04T11:27:21.659846Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.736197Z","title":"Towards making systems forget with machine unlearning","venue":null,"work_id":null,"year":2015},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.736197Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:1ec0c9fd92d32286d694fa7f811c19afae693967630f47375f8bc7ab70777811","observation_id":"95898f54-533a-4a64-93c7-bcc8af62b172","resolution":{"observed_at":"2026-08-04T11:27:21.736197Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.822798Z","title":"Unrolling sgd: Understanding factors influencing machine unlearning","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.822798Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:a7924478e62b200725baa7032deffd11b9bd515b30803831ea7f381c7d3d6044","observation_id":"bc7df84a-8f15-45c4-ba15-5fc45387676d","resolution":{"observed_at":"2026-08-04T11:27:21.822798Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.904258Z","title":"Approximate data deletion from machine learning models","venue":null,"work_id":null,"year":2008},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.904258Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:e0f5eff530c75a72bf6163aca62e8cbf3373ffd28beb0baff06a13fde7e04d2e","observation_id":"427056dd-9a0d-4e4c-8227-871e1f84d102","resolution":{"observed_at":"2026-08-04T11:27:21.904258Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2305.06535","last_updated":"2023-05-11T02:44:29Z","snapshot_observed_at":"2026-08-16T15:33:59.342689Z","submitted_at":"2023-05-11T02:44:29Z","title":"KGA: A General Machine Unlearning Framework Based on Knowledge Gap Alignment","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2305.06535","snapshot_observed_at":"2026-08-04T11:27:22.042626Z","title":"Kga: A general machine unlearning framework based on knowledge gap alignment.arXiv preprint arXiv:2305.06535, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.042626Z"},"links":{"cited_paper":"/paper/2305.06535","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:887fc18bc827781703247d68ecfcf5be096c48dfedbb00fab171cba03c654827","observation_id":"01f61631-3dca-403b-af87-b6b97882fd9d","resolution":{"observed_at":"2026-08-04T11:27:22.042626Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2406.09073","last_updated":"2024-06-13T12:58:00Z","snapshot_observed_at":"2026-08-18T20:10:42.992393Z","submitted_at":"2024-06-13T12:58:00Z","title":"Are we making progress in unlearning? Findings from the first NeurIPS unlearning competition","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2406.09073","snapshot_observed_at":"2026-08-04T11:27:22.168083Z","title":"Are we making progress in unlearning? findings from the first neurips unlearning competition.arXiv preprint arXiv:2406.09073, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.168083Z"},"links":{"cited_paper":"/paper/2406.09073","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:2eced8547f0d3c31a7d222b1cf46cff1822139201cf3b11215ce9445f3656674","observation_id":"fc2b673b-77f5-4978-8004-fff3461956c9","resolution":{"observed_at":"2026-08-04T11:27:22.168083Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.279809Z","title":"Eternal sunshine of the spotless net: Selective forgetting in deep networks","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.279809Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:7db3d52472ba17510af5bd4b10760726bff851678ba7de61fb37de1d26a89149","observation_id":"6c5ee71e-14f3-48c3-89c9-e8946a400926","resolution":{"observed_at":"2026-08-04T11:27:22.279809Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.377946Z","title":"Machine unlearning","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.377946Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:6441f80a91009767744ce5f2d428b626351ecd15a934624de6185c27ee622239","observation_id":"b8d3606d-f684-4666-892b-9267d6c9c27b","resolution":{"observed_at":"2026-08-04T11:27:22.377946Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.494559Z","title":"Model sparsity can simplify machine unlearning.Advances in Neural Information Processing Systems, 36:51584–51605, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.494559Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:08f63e08596b3352394912bf6b8542358b5548f1506feb130a313ff00af097af","observation_id":"14fcf637-2e57-454e-9fba-a047c508a841","resolution":{"observed_at":"2026-08-04T11:27:22.494559Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.12508","last_updated":"2024-04-04T07:45:38Z","snapshot_observed_at":"2026-08-03T17:30:25.153454Z","submitted_at":"2023-10-19T06:17:17Z","title":"SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation","version":5},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.12508","snapshot_observed_at":"2026-08-04T11:27:22.568461Z","title":"Salun: Empowering machine unlearning via gradient-based weight saliency in both image classification and generation.arXiv preprint arXiv:2310.12508, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.568461Z"},"links":{"cited_paper":"/paper/2310.12508","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:a94eed1b9141a2957f8c78f9a17d8b41c47f6f626c41c728899cfe78961564cb","observation_id":"a8242260-85d3-480a-b60d-c4c9d9e20c15","resolution":{"observed_at":"2026-08-04T11:27:22.568461Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.581625Z","title":"Towards un- bounded machine unlearning.Advances in neural information processing systems, 36:1957– 1987, 2023","venue":null,"work_id":null,"year":1957},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.581625Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:c42a59dc801ddd6db3d453f650e15cb30709555ad2674098c7f13fceb8a368b0","observation_id":"28d5263f-751c-465b-aa6f-563363a0a433","resolution":{"observed_at":"2026-08-04T11:27:22.581625Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.592812Z","title":"Making ai forget you: Data deletion in machine learning.Advances in neural information processing systems, 32, 2019","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.592812Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:b1f4b07f29e0e1a39f9ba7f2a99d2f0709eb2e7fbbd37fecdb7ebeea0e5b2d12","observation_id":"de3d0565-6a7d-4220-a9f1-38fce8a050b0","resolution":{"observed_at":"2026-08-04T11:27:22.592812Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.604414Z","title":"Erasing concepts from diffusion models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.604414Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:a176514ddcedda9216e71e3c2c649e80ad7b4af73b008881c75b12a4721dc1a2","observation_id":"6dcd6cc9-c133-43fc-94dc-6d83c3093a42","resolution":{"observed_at":"2026-08-04T11:27:22.604414Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.611862Z","title":"Forget-me- not: Learning to forget in text-to-image diffusion models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.611862Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:d1404d665882e748d0f197c630d4b93ab854b794504540a1eb8a983d78e49956","observation_id":"1760d722-3b95-469a-90ab-4ab668aa86b4","resolution":{"observed_at":"2026-08-04T11:27:22.611862Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.617452Z","title":"Fast federated machine unlearning with nonlinear functional theory","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.617452Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:eb35f6fb85bce5e9943e14d57daf5dc40352c0b97c31507b8e8aeef37e2a43f3","observation_id":"e37871ed-e77b-42af-a9ce-72061651491b","resolution":{"observed_at":"2026-08-04T11:27:22.617452Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.627175Z","title":"Federated unlearning with gradient descent and conflict mitigation","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.627175Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:5fe250210fe0a0f4d34bd4ebc2e7d7d7357810f489c9d3466c0808a0edded835","observation_id":"b68d6268-6e49-4123-9d44-72b01997e618","resolution":{"observed_at":"2026-08-04T11:27:22.627175Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.632281Z","title":"Efficient model updates for approximate un- learning of graph-structured data","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.632281Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:2e3704ea9677a0568b7d2da832e3d0cf21c80a4c0848cfcb3ee51930dc9ef5f7","observation_id":"305b3c49-1a8a-45af-8cfe-7bc3962e438d","resolution":{"observed_at":"2026-08-04T11:27:22.632281Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.641600Z","title":"Certified edge unlearning for graph neural networks","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.641600Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:22a45bd18e0a6ec507b8c8c632897f19a8edabcda675c6d86cbfed02ef359264","observation_id":"31ba6a7a-0881-4468-9f1e-af7642a4c274","resolution":{"observed_at":"2026-08-04T11:27:22.641600Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.654558Z","title":"Machine unlearning for recommendation systems: An insight","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.654558Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:1df0ebebfe98d12b429b418b64608a2902e738352b29e81d8ef3cea64364253b","observation_id":"8983a749-7907-49c8-b55f-673239cb3a4b","resolution":{"observed_at":"2026-08-04T11:27:22.654558Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.669615Z","title":"Simplicity prevails: Rethinking negative preference optimization for llm unlearning.arXiv preprint arXiv:2410.07163, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.669615Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:7ef6e1d8e6e723db91233203c828e4a6d2650d3ea95f56f4a544dc012810e693","observation_id":"8a4c8eae-a8ab-4b0b-a1fc-2f36196842e4","resolution":{"observed_at":"2026-08-04T11:27:22.669615Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2409.13474","last_updated":"2024-12-17T17:45:07Z","snapshot_observed_at":"2026-08-16T13:16:52.832959Z","submitted_at":"2024-09-20T13:05:07Z","title":"Alternate Preference Optimization for Unlearning Factual Knowledge in Large Language Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2409.13474","snapshot_observed_at":"2026-08-04T11:27:22.676125Z","title":"Alternate preference optimization for unlearning factual knowledge in large language models.arXiv preprint arXiv:2409.13474, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.676125Z"},"links":{"cited_paper":"/paper/2409.13474","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:04cced1a9d6279ec48ba341842e75e469b991dcef5e29581786293f152674f33","observation_id":"077c96d1-f879-4768-8784-4243f5c9211f","resolution":{"observed_at":"2026-08-04T11:27:22.676125Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2404.18239","last_updated":"2024-06-24T20:24:53Z","snapshot_observed_at":"2026-08-16T13:57:11.558894Z","submitted_at":"2024-04-28T16:31:32Z","title":"SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2404.18239","snapshot_observed_at":"2026-08-04T11:27:22.687906Z","title":"Soul: Unlocking the power of second-order optimization for llm unlearning.arXiv preprint arXiv:2404.18239, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.687906Z"},"links":{"cited_paper":"/paper/2404.18239","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:181d4a4eb4b2bc80f4b9b5fa98ca069576329a424852027572d3c1750315a753","observation_id":"0ba8b6cb-9b38-40dc-8b5f-19cabd0453e1","resolution":{"observed_at":"2026-08-04T11:27:22.687906Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.699805Z","title":"Can bad teaching induce forgetting? unlearning in deep networks using an incompetent teacher","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.699805Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:801750d5289ebe7fe769636f7aed9a1a04410f4cea2b36f0b48f356430063885","observation_id":"35d2aee4-66cd-4d6d-962a-6095ba727c2c","resolution":{"observed_at":"2026-08-04T11:27:22.699805Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.706659Z","title":"Reversing the forget-retain objectives: An efficient llm unlearning framework from logit difference.Advances in Neural Information Processing Systems, 37:12581–12611, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.706659Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:d522c4c0b471d42e566ab7b03f03313463b074baf385e6f3c08ad1979c4df294","observation_id":"622fe18a-75f4-44b6-958d-688eaa748e6b","resolution":{"observed_at":"2026-08-04T11:27:22.706659Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.20150","last_updated":"2023-10-31T03:35:59Z","snapshot_observed_at":"2026-08-16T14:47:23.256137Z","submitted_at":"2023-10-31T03:35:59Z","title":"Unlearn What You Want to Forget: Efficient Unlearning for LLMs","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.20150","snapshot_observed_at":"2026-08-04T11:27:22.713758Z","title":"Unlearn what you want to forget: Efficient unlearning for llms","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":39,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.713758Z"},"links":{"cited_paper":"/paper/2310.20150","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:8306ffa4a913708ed4254b57935f14056aa88148240c5e9a4835852bec4b5afb","observation_id":"7ca5c113-7c77-49a6-897f-1dcb16bc08ec","resolution":{"observed_at":"2026-08-04T11:27:22.713758Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2402.15159","last_updated":"2024-05-30T15:44:51Z","snapshot_observed_at":"2026-08-17T08:25:32.062258Z","submitted_at":"2024-02-23T07:43:26Z","title":"Machine Unlearning of Pre-trained Large Language Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2402.15159","snapshot_observed_at":"2026-08-04T11:27:22.723583Z","title":"Machine unlearning of pre-trained large language models.arXiv preprint arXiv:2402.15159, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":40,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.723583Z"},"links":{"cited_paper":"/paper/2402.15159","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:b36e3c2dbe44c9ff1d16ec9c5dcfe8a6476ae7631ee725aaefdc150b2354e6f4","observation_id":"63d4d8b0-1a9f-4c96-91b4-2a8d6a87eda1","resolution":{"observed_at":"2026-08-04T11:27:22.723583Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2309.11852","last_updated":"2024-03-02T08:50:40Z","snapshot_observed_at":"2026-08-20T13:15:50.076184Z","submitted_at":"2023-09-21T07:49:55Z","title":"Knowledge Sanitization of Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2309.11852","snapshot_observed_at":"2026-08-04T11:27:22.736643Z","title":"Knowledge sanitization of large language models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":41,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.736643Z"},"links":{"cited_paper":"/paper/2309.11852","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:f763c3a25c0c3b3fc1f6b73a50d68bb37758390161d91003479579d0d7e1e8a4","observation_id":"1aaa2258-65ea-48e9-9abe-89fa5edc6385","resolution":{"observed_at":"2026-08-04T11:27:22.736643Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2407.16997","last_updated":"2024-10-07T03:41:57Z","snapshot_observed_at":"2026-08-19T13:49:36.010591Z","submitted_at":"2024-07-24T04:39:24Z","title":"Revisiting Who's Harry Potter: Towards Targeted Unlearning from a Causal Intervention Perspective","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2407.16997","snapshot_observed_at":"2026-08-04T11:27:22.749242Z","title":"Revisiting who’s harry pot- ter: Towards targeted unlearning from a causal intervention perspective.arXiv preprint arXiv:2407.16997, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":42,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.749242Z"},"links":{"cited_paper":"/paper/2407.16997","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:4649f1ca158afca9cd932f0268ae2a751fbe0337935ab464f321a643129c6499","observation_id":"61cb8d93-3765-4446-b6e2-f45170fd1cbb","resolution":{"observed_at":"2026-08-04T11:27:22.749242Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2403.03329","last_updated":"2024-06-11T15:47:39Z","snapshot_observed_at":"2026-08-20T00:34:57.831784Z","submitted_at":"2024-03-05T21:19:06Z","title":"Guardrail Baselines for Unlearning in LLMs","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2403.03329","snapshot_observed_at":"2026-08-04T11:27:22.757191Z","title":"Guardrail baselines for unlearning in llms.arXiv preprint arXiv:2403.03329, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":43,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.757191Z"},"links":{"cited_paper":"/paper/2403.03329","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:a7d59a812ea63a873d46830e27605ac519401d59b94eccd96ea60037041c128b","observation_id":"21f9fa8b-8f0f-490e-be1f-e310b73050a3","resolution":{"observed_at":"2026-08-04T11:27:22.757191Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.07579","last_updated":"2024-06-06T06:31:08Z","snapshot_observed_at":"2026-08-21T22:14:38.761302Z","submitted_at":"2023-10-11T15:19:31Z","title":"In-Context Unlearning: Language Models as Few Shot Unlearners","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.07579","snapshot_observed_at":"2026-08-04T11:27:22.765817Z","title":"In-context unlearning: Language models as few shot unlearners.arXiv preprint arXiv:2310.07579, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":44,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.765817Z"},"links":{"cited_paper":"/paper/2310.07579","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:f70913087763da95f56bc5f873266f42156b7cebfef9178562be2e33d99e799b","observation_id":"65baeb04-f295-4b0d-9eb6-d9efbf38b680","resolution":{"observed_at":"2026-08-04T11:27:22.765817Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.777586Z","title":"Training language models to follow instructions with human feedback.Advances in neural information processing systems, 35:27730–27744, 2022","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":45,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.777586Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:e7d668f22c26031a572eaa234b5aff652c49e451e2f43e9d1bc2d846a1f69707","observation_id":"6f10c97d-7c95-4ac6-93b1-d128acc176f5","resolution":{"observed_at":"2026-08-04T11:27:22.777586Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2403.07691","last_updated":"2024-03-14T07:47:08Z","snapshot_observed_at":"2026-08-16T14:56:35.674907Z","submitted_at":"2024-03-12T14:34:08Z","title":"ORPO: Monolithic Preference Optimization without Reference Model","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2403.07691","snapshot_observed_at":"2026-08-04T11:27:22.792715Z","title":"Orpo: Monolithic preference optimization without reference model.arXiv preprint arXiv:2403.07691, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":46,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.792715Z"},"links":{"cited_paper":"/paper/2403.07691","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:672849bcf1d53ab79ff484d9f45b362a6d19478058cae590950d98ccab04cd30","observation_id":"82f0f4b3-e1e8-4975-b743-c72a044b7cf6","resolution":{"observed_at":"2026-08-04T11:27:22.792715Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.800337Z","title":"Simpo: Simple preference optimization with a reference-free reward.Advances in Neural Information Processing Systems, 37:124198–124235, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":47,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.800337Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:3ab760d25cc9f797cb3d03520b5b2191d6f6e459894fe4c0b081b2d5f6ae497b","observation_id":"442eb49d-2fa8-46e9-92b5-db9aec14a9dd","resolution":{"observed_at":"2026-08-04T11:27:22.800337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2309.16240","last_updated":"2023-09-28T08:29:44Z","snapshot_observed_at":"2026-08-19T00:14:01.230491Z","submitted_at":"2023-09-28T08:29:44Z","title":"Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2309.16240","snapshot_observed_at":"2026-08-04T11:27:22.818786Z","title":"Beyond reverse kl: Generalizing direct preference optimization with diverse divergence constraints.arXiv preprint arXiv:2309.16240, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":48,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.818786Z"},"links":{"cited_paper":"/paper/2309.16240","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:ac5e5d63f40cb9377076d3615e0dbb5f8dd7bb13b72b09951b7a41bdd6d5d740","observation_id":"951663bd-9add-4f69-9910-0ce7a4bcc7e2","resolution":{"observed_at":"2026-08-04T11:27:22.818786Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.832649Z","title":"A general theoretical paradigm to understand learning from human preferences","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":49,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.832649Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:db50171b3f8381cc6e2c49fa2a99ff22414bb4be1e4ebdc3b6f599937108b3f4","observation_id":"3bdb80e2-0b17-4b4f-91f7-3dd1fcb56502","resolution":{"observed_at":"2026-08-04T11:27:22.832649Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.843643Z","title":"Generalizing offline alignment theoretical paradigm with diverse divergence constraints","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":50,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.843643Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:a0d93471a89b25cc32b45143ad2dbc15e8091dbda93cfaa6099cb62e6b3578ea","observation_id":"c66c64f5-2641-4af1-ac1f-b91dc604551b","resolution":{"observed_at":"2026-08-04T11:27:22.843643Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2404.11999","last_updated":"2024-08-30T03:39:57Z","snapshot_observed_at":"2026-08-18T22:30:23.108990Z","submitted_at":"2024-04-18T08:49:38Z","title":"Token-level Direct Preference Optimization","version":5},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2404.11999","snapshot_observed_at":"2026-08-04T11:27:22.852709Z","title":"Token- level direct preference optimization.arXiv preprint arXiv:2404.11999, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":51,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.852709Z"},"links":{"cited_paper":"/paper/2404.11999","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:2aa7cccb3d1bd85732c5458408e9ea66fd8d80e023e952490c473558831d8c72","observation_id":"39310ff7-28d9-4fea-889c-7381e850af02","resolution":{"observed_at":"2026-08-04T11:27:22.852709Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2210.15097","last_updated":"2023-07-10T06:08:55Z","snapshot_observed_at":"2026-08-18T17:10:17.670255Z","submitted_at":"2022-10-27T00:58:21Z","title":"Contrastive Decoding: Open-ended Text Generation as Optimization","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2210.15097","snapshot_observed_at":"2026-08-04T11:27:22.860151Z","title":"Contrastive decoding: Open-ended text generation as optimization.arXiv preprint arXiv:2210.15097, 2022","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":52,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.860151Z"},"links":{"cited_paper":"/paper/2210.15097","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:3102e4b81eb7046480abf49e506df484d9087d0d9981ec1ff924a6bee8c61882","observation_id":"a09e114b-10a6-446d-82b4-e6338fb04a64","resolution":{"observed_at":"2026-08-04T11:27:22.860151Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2309.03883","last_updated":"2024-03-11T02:01:09Z","snapshot_observed_at":"2026-08-06T11:43:44.726916Z","submitted_at":"2023-09-07T17:45:31Z","title":"DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2309.03883","snapshot_observed_at":"2026-08-04T11:27:22.867379Z","title":"Dola: Decoding by contrasting layers improves factuality in large language models.arXiv preprint arXiv:2309.03883, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":53,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.867379Z"},"links":{"cited_paper":"/paper/2309.03883","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:d4e1847da77a62980735ed26b575a361be87d455b5fad32be09883f9ff96a56e","observation_id":"533cbf42-75a5-410c-8900-f73a7152f965","resolution":{"observed_at":"2026-08-04T11:27:22.867379Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2105.03023","last_updated":"2021-06-03T05:26:11Z","snapshot_observed_at":"2026-08-20T16:15:19.632317Z","submitted_at":"2021-05-07T01:19:38Z","title":"DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2105.03023","snapshot_observed_at":"2026-08-04T11:27:22.875112Z","title":"Dexperts: Decoding-time controlled text generation with experts and anti-experts.arXiv preprint arXiv:2105.03023, 2021","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":54,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.875112Z"},"links":{"cited_paper":"/paper/2105.03023","citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:eaced6f4f20df4584cd93544116ff748542d4702aa18138f956c4d493dadc38a","observation_id":"d983ffd8-7f3c-4346-8ab4-2e721874330c","resolution":{"observed_at":"2026-08-04T11:27:22.875112Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.883918Z","title":"I don’t know","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":56,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.883918Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:738ae834538c1d0641f8b3eaa5788836065755eb32ed08cba9df647a5b0fa1a7","observation_id":"b5b445b7-29a0-413b-bd58-efd3679a9c9f","resolution":{"observed_at":"2026-08-04T11:27:22.883918Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.893044Z","title":"It is set to the log-probability corresponding to the k-th rank when tokens are sorted by log-probability in descending order, wherek=max(1,⌊p k · |V|⌋)","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":57,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.893044Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:1644a7932fff4f2d243b937437d68787a1b3ccb61534e8167b595963bea0b663","observation_id":"84bc9d9d-87d3-4783-b656-deb234dd806f","resolution":{"observed_at":"2026-08-04T11:27:22.893044Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.906273Z","title":"The final threshold used for filtering is the minimum of these two: τ=min(τ k,τ rel)","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":58,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.906273Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:da7b62aa0c1a6b0fe3cb995c3e0fa7d61069dfe26d61bc3413596ef7ca242e13","observation_id":"aa2f7269-327e-4504-af77-d181ba846109","resolution":{"observed_at":"2026-08-04T11:27:22.906273Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.918310Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":59,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.918310Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:5e2b68e9b7ad9328343a34e84e19f437dd67a6814e7f168d49bb33d75e0b377e","observation_id":"e7987f28-a8fd-44db-8d94-22865e0676e4","resolution":{"observed_at":"2026-08-04T11:27:22.918310Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.927468Z","title":"D.4.2 Evaluation Metrics MUSE evaluates unlearning across six criteria","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":60,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.927468Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:aa95ae1bbfa71c036478fbc6382a1776aac3a01acb11b2d89f2b09b18b77a10b","observation_id":"8636b27f-491c-4d6e-b016-55fb9c0ea1a1","resolution":{"observed_at":"2026-08-04T11:27:22.927468Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.940525Z","title":"Quantified by VerbMem-f, which measures the ROUGE-L F1 score between model-generated continuations and true continuations fromD forget","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":61,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.940525Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:6ac8e26455bea6c8d762b52e3453dff0a0592cb40488124510524c96a77c4008","observation_id":"ab678436-6c6b-44db-ae6e-36a55acfb520","resolution":{"observed_at":"2026-08-04T11:27:22.940525Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.952142Z","title":"Quantified by KnowMem-f, averaging ROUGE scores between model answers and ground-truth answers for QA pairs derived fromD forget","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":62,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.952142Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:1a7207609a328e938e6509a843dfff71a43a03ff0898a2cd994c8a8dc9368685","observation_id":"58365d12-b059-44af-af1f-d0c1ab031d7f","resolution":{"observed_at":"2026-08-04T11:27:22.952142Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.962003Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":63,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.962003Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:75c4fbac66f57c02c33efc8a5e22116fa28aef7128c29bef55247ae59478d53a","observation_id":"c48ed92a-a869-4276-8ef7-bd3726db3fad","resolution":{"observed_at":"2026-08-04T11:27:22.962003Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.970398Z","title":"This is typically measured using theKnowMem-rmetric applied toD retain: KnowMem-r(f unlearn,D retain)","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":64,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.970398Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:185a9ce2db96e3a984939aeb33d3f0fe77b7c9c53fea9f4040b51b24cac623a3","observation_id":"e7eccd9b-9864-4885-b52c-75c1e912e191","resolution":{"observed_at":"2026-08-04T11:27:22.970398Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.979127Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":65,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.979127Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:7037c8ef5178e9567a48ba17b0db1535ceaf71072471a039767b29253e4eeafa","observation_id":"86721fe3-4bff-478b-9026-c4c8f0199470","resolution":{"observed_at":"2026-08-04T11:27:22.979127Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.987509Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":66,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.987509Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:30965532f2aba5404ff92e33a947e77a6f8781d8e5953d74a340085bfb2c7bfd","observation_id":"3076dcc5-c606-4456-bc56-1cb4ae2c3002","resolution":{"observed_at":"2026-08-04T11:27:22.987509Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:22.997014Z","title":"The combined objective is then expressed asL=L DiPO-f(θ) +λLDiPO-r(θ)","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":67,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:22.997014Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:b33218224f000778bfd0e09b07106477ad1eaa1d453d7818a9f3751cd601ee5b","observation_id":"3097a34c-01b2-44dc-8ac5-94b0e7e4b5ca","resolution":{"observed_at":"2026-08-04T11:27:22.997014Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:23.007306Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":68,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:23.007306Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:9c99b0e850c2db89ee0e73ef183255ab7b8af99dd11a60ce464fe4e361ef0dfa","observation_id":"bce5f7da-24f5-4e68-babc-27420bf57f32","resolution":{"observed_at":"2026-08-04T11:27:23.007306Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:23.012495Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":69,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:23.012495Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:513a577e76d12d4a0e9aaee5f0b713eaca2363923930b81f8fb8bd347cc27fe6","observation_id":"0bd42952-c11d-41e6-a935-0147f48d2034","resolution":{"observed_at":"2026-08-04T11:27:23.012495Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:23.020889Z","title":"Additionally, we discuss DiPO-Forget (using only LDiPO-f without any retain loss)","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":70,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:23.020889Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:31648753095c7d3dea988e877aa8bbe190a6a0f88187d1c4a5a18eeee8015bf6","observation_id":"086c3d83-9bac-4b22-9317-29964f111ebf","resolution":{"observed_at":"2026-08-04T11:27:23.020889Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T11:27:21.088646Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning","version":2},"reference_index":2018,"source":"pdf_text","source_observed_at":"2026-08-04T11:27:21.088646Z"},"links":{"citing_paper":"/paper/2510.04773"},"observation_digest":"sha256:3be2ac37dbd617d0de57afd849d350696cfa309b61e92c9309c6358376eee571","observation_id":"51f2608d-4c44-46eb-80b4-5544db1bae1f","resolution":{"observed_at":"2026-08-04T11:27:21.088646Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2510.04773","last_updated":"2026-06-24T11:41:35Z","latest_version":2,"primary_category":"cs.LG","snapshot_observed_at":"2026-08-18T15:49:22.616263Z","submitted_at":"2025-10-06T12:49:00Z","title":"Distribution Preference Optimization: A Fine-grained Perspective for LLM Unlearning"},"reference_resolution":{"displayed":70,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":70,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":70},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-23T06:30:58.430688+00:00","source":"crossref"},{"observed_at":"2026-08-23T06:30:53.778098+00:00","source":"retraction_watch"}],"thesis":"As of 23 August 2026, this Paper Citation Record lists 70 of 70 outbound references and 2 inbound Pith citation observations for arXiv:2510.04773."}