{"as_of":"2026-08-09T23:12:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:8e1ed6722defd1e73992c251c038d08128b3c0e56d02cd53c761b69df1d06946","coverage":[{"denominator":19,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":19,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-03T18:28:33.174280Z","state":"measured"},{"denominator":19,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":19,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-09T06:31:02.800959+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2512.05254/citation-record","integrity":"/paper/2512.05254/integrity","json":"/paper/2512.05254/citation-record.json","paper":"/paper/2512.05254"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-03T18:28:31.089792Z","title":"Deep learning with differential privacy","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.089792Z"},"links":{"citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:038aeebd63f0992059ff2776c185363278ccd46f5cd475054b75e589e3a52160","observation_id":"4b3fee4e-ef15-4afa-91e5-6a8ed203dda9","resolution":{"observed_at":"2026-08-03T18:28:31.089792Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2008.03703","last_updated":"2020-08-09T10:12:28Z","snapshot_observed_at":"2026-08-09T03:42:42.714546Z","submitted_at":"2020-08-09T10:12:28Z","title":"What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimation","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2008.03703","snapshot_observed_at":"2026-08-03T18:28:31.768107Z","title":"Ryan Giordano, Will Stephenson, Runjing Liu, Michael I","venue":null,"work_id":null,"year":2008},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.768107Z"},"links":{"cited_paper":"/paper/2008.03703","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:1509e1e875a5b76573e6c3a5a6f471a77f4c842d713497329f65d580e8617edc","observation_id":"7305edd2-54f1-4a42-9cab-c644badcfdbe","resolution":{"observed_at":"2026-08-03T18:28:31.768107Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1806.00550","last_updated":"2020-02-07T17:18:46Z","snapshot_observed_at":"2026-08-08T21:00:07.677078Z","submitted_at":"2018-06-01T21:48:44Z","title":"A Swiss Army Infinitesimal Jackknife","version":5},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1806.00550","snapshot_observed_at":"2026-08-03T18:28:31.938971Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.938971Z"},"links":{"cited_paper":"/paper/1806.00550","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:74b3b19e8e54f8dad0e6727b7a82f4e84fa9fb700b6834a0f578a9bcecb8c2ad","observation_id":"52757022-dcf5-4f7b-a2b9-a84d011d2908","resolution":{"observed_at":"2026-08-03T18:28:31.938971Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2308.03296","last_updated":"2023-08-07T04:47:42Z","snapshot_observed_at":"2026-08-09T11:08:40.708986Z","submitted_at":"2023-08-07T04:47:42Z","title":"Studying Large Language Model Generalization with Influence Functions","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2308.03296","snapshot_observed_at":"2026-08-03T18:28:32.058194Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.058194Z"},"links":{"cited_paper":"/paper/2308.03296","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:ddfd094c43f66552c370194fe39baceae220a75f2e352dfda6863d7b11e9489e","observation_id":"1a25b215-16af-485a-a89d-ae9596289537","resolution":{"observed_at":"2026-08-03T18:28:32.058194Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1703.04730","last_updated":"2020-12-29T22:40:43Z","snapshot_observed_at":"2026-08-09T22:46:27.354385Z","submitted_at":"2017-03-14T21:07:01Z","title":"Understanding Black-box Predictions via Influence Functions","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1703.04730","snapshot_observed_at":"2026-08-03T18:28:32.268921Z","title":"Alex Krizhevsky, Geoffrey Hinton, et al","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.268921Z"},"links":{"cited_paper":"/paper/1703.04730","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:63a966074558df9dfd0149658731f5a86624f832670e65103bbae3d7ce033733","observation_id":"0a748625-897f-4672-8df3-7e4b3380310d","resolution":{"observed_at":"2026-08-03T18:28:32.268921Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2002.08484","last_updated":"2020-11-14T18:47:35Z","snapshot_observed_at":"2026-08-09T03:58:35.037012Z","submitted_at":"2020-02-19T22:40:32Z","title":"Estimating Training Data Influence by Tracing Gradient Descent","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2002.08484","snapshot_observed_at":"2026-08-03T18:28:32.507165Z","title":"P Rajpurkar","venue":null,"work_id":null,"year":2002},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.507165Z"},"links":{"cited_paper":"/paper/2002.08484","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:0bfebdf0a536cdb9809d8fc4d7a4d5c6ba2d2500652ea670078e4db3905afcda","observation_id":"446f04a7-9d6d-434d-93ff-edcac189c2ed","resolution":{"observed_at":"2026-08-03T18:28:32.507165Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2307.03941","last_updated":"2024-06-05T01:14:19Z","snapshot_observed_at":"2026-07-06T15:51:43.130122Z","submitted_at":"2023-07-08T09:28:50Z","title":"Right to be Forgotten in the Era of Large Language Models: Implications, Challenges, and Solutions","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2307.03941","snapshot_observed_at":"2026-08-03T18:28:32.926699Z","title":"URLhttps://doi.org/10","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.926699Z"},"links":{"cited_paper":"/paper/2307.03941","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:323daa21491d6f472f212b9ff3c40abdfc771cecb9524ba8898c7ba9137bb599","observation_id":"cdd20980-fec3-4ebc-aa54-e52fd9bf09b4","resolution":{"observed_at":"2026-08-03T18:28:32.926699Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1908.10654","last_updated":"2020-02-04T17:58:24Z","snapshot_observed_at":"2026-08-05T03:49:23.935968Z","submitted_at":"2019-08-28T11:40:51Z","title":"CASIA-SURF: A Large-scale Multi-modal Benchmark for Face Anti-spoofing","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1908.10654","snapshot_observed_at":"2026-08-03T18:28:33.063415Z","title":null,"venue":null,"work_id":null,"year":1908},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:33.063415Z"},"links":{"cited_paper":"/paper/1908.10654","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:5913f78b42c4f48556ead9581baa7e3e44d746cc98acb2914e34cd39da6c2898","observation_id":"b029dc7b-9d30-41b1-9dcc-9f5ea8de8577","resolution":{"observed_at":"2026-08-03T18:28:33.063415Z","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":"2026-08-03T18:28:33.174280Z","title":"epochs\" and the learning rate of","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":128,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:33.174280Z"},"links":{"citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:00eb87b903bfed2ea60c584eb57ac9fbd949b724f8d6a1919ace34675a2f6e17","observation_id":"12df5b54-2ade-43c7-a9e7-f83d6fbe297d","resolution":{"observed_at":"2026-08-03T18:28:33.174280Z","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":"2026-08-03T18:28:32.173941Z","title":"URLhttps: //www.tandfonline.com/doi/abs/10.1080/01621459.1974.10482962","venue":null,"work_id":null,"year":1974},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":1974,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.173941Z"},"links":{"citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:9900e77656565f8a670bfaaecdb866494a88ace2a96aa73dd2f38bf37c27880c","observation_id":"5c380d18-0156-42cb-8579-ed65883efdcf","resolution":{"observed_at":"2026-08-03T18:28:32.173941Z","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":"2026-08-03T18:28:31.561149Z","title":"doi: 10.1561/0400000042","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2014,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.561149Z"},"links":{"citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:27f5b72e129c3fd2c9889dc59922c13141b43a2c82a7c46ebaacc8f3c200529a","observation_id":"bec5cbbb-6055-4e7e-ab44-c4a2ecb80628","resolution":{"observed_at":"2026-08-03T18:28:31.561149Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1810.04805","last_updated":"2019-05-24T20:37:26Z","snapshot_observed_at":"2026-07-30T09:12:38.100527Z","submitted_at":"2018-10-11T00:50:01Z","title":"BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1810.04805","snapshot_observed_at":"2026-08-03T18:28:31.418282Z","title":"Jacob Devlin","venue":null,"work_id":null,"year":2015},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2015,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.418282Z"},"links":{"cited_paper":"/paper/1810.04805","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:c55df7b88d2bb1d19eb89f3a1a9c90859287d4485d8fc9be4cdc15358e9fd0fb","observation_id":"23b47b72-c6e5-415c-962f-eacec570cd13","resolution":{"observed_at":"2026-08-03T18:28:31.418282Z","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":"2026-08-03T18:28:31.136601Z","title":"URLhttp: //dx.doi.org/10.1145/2976749.2978318","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2016,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.136601Z"},"links":{"citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:bf38ac967624864357475937218357725243fbf55fa81b6761aa555f0352b81f","observation_id":"6eb60072-0966-4029-8160-f64f24685f30","resolution":{"observed_at":"2026-08-03T18:28:31.136601Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1610.05820","last_updated":"2017-03-31T22:17:07Z","snapshot_observed_at":"2026-07-06T05:15:08.120372Z","submitted_at":"2016-10-18T22:38:33Z","title":"Membership Inference Attacks against Machine Learning Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1610.05820","snapshot_observed_at":"2026-08-03T18:28:32.659749Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2017,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.659749Z"},"links":{"cited_paper":"/paper/1610.05820","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:9060e369e692e1b956bee1fc220c5b65afae638fabaaf33588883ba9586ddd97","observation_id":"be2211e6-435c-42df-acc3-f05a982c2ee5","resolution":{"observed_at":"2026-08-03T18:28:32.659749Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1912.03817","last_updated":"2020-12-15T05:39:28Z","snapshot_observed_at":"2026-08-09T19:27:32.079920Z","submitted_at":"2019-12-09T02:16:53Z","title":"Machine Unlearning","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1912.03817","snapshot_observed_at":"2026-08-03T18:28:31.230309Z","title":"Tamara Broderick, Ryan Giordano, and Rachael Meager","venue":null,"work_id":null,"year":1912},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2020,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.230309Z"},"links":{"cited_paper":"/paper/1912.03817","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:f2eb8e61c140e2c5aed88f0039cedf137b91220eb6b2a485d514576bc348d29b","observation_id":"1f3a8acc-3023-498f-9f53-83d366028570","resolution":{"observed_at":"2026-08-03T18:28:31.230309Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1906.05271","last_updated":"2021-01-10T07:19:48Z","snapshot_observed_at":"2026-08-07T14:02:44.326094Z","submitted_at":"2019-06-12T17:53:25Z","title":"Does Learning Require Memorization? A Short Tale about a Long Tail","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1906.05271","snapshot_observed_at":"2026-08-03T18:28:31.665232Z","title":"Vitaly Feldman and Chiyuan Zhang","venue":null,"work_id":null,"year":1906},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2021,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.665232Z"},"links":{"cited_paper":"/paper/1906.05271","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:10602e6c35ce9c59dc00599315b4f298ac336c36ecdbedb0384406c8d62d8d08","observation_id":"bc8b8263-8b9e-4f95-a9e4-f34468f99547","resolution":{"observed_at":"2026-08-03T18:28:31.665232Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2011.14999","last_updated":"2023-07-19T17:09:07Z","snapshot_observed_at":"2026-08-09T09:47:51.181800Z","submitted_at":"2020-11-30T17:05:48Z","title":"An Automatic Finite-Sample Robustness Metric: When Can Dropping a Little Data Make a Big Difference?","version":5},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2011.14999","snapshot_observed_at":"2026-08-03T18:28:31.326947Z","title":"Yinzhi Cao and Junfeng Yang","venue":null,"work_id":null,"year":2011},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2023,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:31.326947Z"},"links":{"cited_paper":"/paper/2011.14999","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:35f0cd997a672eb54ae051598eac7582177f6987c3ea2a7604f0761a761a7e1d","observation_id":"88957db3-c618-4656-ae67-1148a55eb222","resolution":{"observed_at":"2026-08-03T18:28:31.326947Z","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":"2026-08-03T19:43:59.941345Z","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":"2026-08-03T18:28:32.783687Z","title":"Mengzhou Xia, Sadhika Malladi, Suchin Gururangan, Sanjeev Arora, and Danqi Chen","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2024,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.783687Z"},"links":{"cited_paper":"/paper/2406.09073","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:bd8467fd092b37a618e0fd0d0a006376ddb00a1685786a5354d0294013f3dac6","observation_id":"613ad083-3961-419f-8842-ef4b696394e4","resolution":{"observed_at":"2026-08-03T18:28:32.783687Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2501.16614","last_updated":"2025-01-28T01:19:07Z","snapshot_observed_at":"2026-08-09T21:21:27.599980Z","submitted_at":"2025-01-28T01:19:07Z","title":"FUNU: Boosting Machine Unlearning Efficiency by Filtering Unnecessary Unlearning","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2501.16614","snapshot_observed_at":"2026-08-03T18:28:32.408652Z","title":"Ken Ziyu Liu","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs","version":2},"reference_index":2025,"source":"pdf_text","source_observed_at":"2026-08-03T18:28:32.408652Z"},"links":{"cited_paper":"/paper/2501.16614","citing_paper":"/paper/2512.05254"},"observation_digest":"sha256:0e566366cc2a6c37b6b8b5ad255357ed8c4d7fd6d03a47a7b6551c6cd83f0112","observation_id":"d7f163fe-f2a6-4d42-88b5-8549dce692d3","resolution":{"observed_at":"2026-08-03T18:28:32.408652Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2512.05254","last_updated":"2026-07-29T23:30:29Z","latest_version":2,"primary_category":"cs.LG","snapshot_observed_at":"2026-08-09T15:32:48.696471Z","submitted_at":"2025-12-04T21:10:31Z","title":"When unlearning is free: leveraging low influence points to reduce computational costs"},"reference_resolution":{"displayed":19,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":19,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":19},"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":"2026-08-09T06:31:02.800959+00:00","source":"crossref"},{"observed_at":"2026-08-09T06:30:57.326959+00:00","source":"retraction_watch"}],"thesis":"As of 9 August 2026, this Paper Citation Record lists 19 of 19 outbound references and 0 inbound Pith citation observations for arXiv:2512.05254."}