{"as_of":"2026-08-15T03:43:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:a9e4accb05ef1b547373aec35f21bbe0e76d6c0f1145f80a2da2458082118e7a","coverage":[{"denominator":68,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":68,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-08T20:08:57.514936Z","state":"measured"},{"denominator":68,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":68,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-14T06:32:32.682623+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/2502.05159/citation-record","integrity":"/paper/2502.05159/integrity","json":"/paper/2502.05159/citation-record.json","paper":"/paper/2502.05159"},"outbound":[{"citation":{"cited_paper":{"arxiv_id":"2412.06619","last_updated":"2024-12-09T16:13:17Z","snapshot_observed_at":"2026-08-14T17:18:15.856205Z","submitted_at":"2024-12-09T16:13:17Z","title":"Copyright-Protected Language Generation via Adaptive Model Fusion","version":1},"cited_work":{"arxiv_id":"2412.06619","doi":null,"metadata_source":"pith","pith_arxiv_id":"2412.06619","snapshot_observed_at":"2026-08-08T20:08:57.949206Z","title":"Copyright-Protected Language Generation via Adaptive Model Fusion","venue":"cs.LG","work_id":"4c59d8fe-2377-4460-b3cf-79819b27584c","year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":1,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.247629Z"},"links":{"cited_paper":"/paper/2412.06619","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:2e1fad3cce94101d185ce7cdf27d0703f4fcc597c3f1b4e13920978454f78896","observation_id":"fc06a381-c6de-4988-94eb-a172d01e7931","resolution":{"observed_at":"2026-08-08T20:08:57.953888Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:57.253008Z","title":"Deep learning with differential privacy","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":2,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.253008Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:8d5ece1ec54b52d771b85bf0da16ae5f315be57d6083a157afc27ce183854697","observation_id":"6fbe8091-c76a-4f29-993d-7f213dd0d3e1","resolution":{"observed_at":"2026-08-08T20:08:57.253008Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2303.08774","last_updated":"2024-03-04T06:01:33Z","snapshot_observed_at":"2026-08-07T07:30:12.213965Z","submitted_at":"2023-03-15T17:15:04Z","title":"GPT-4 Technical Report","version":6},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2303.08774","snapshot_observed_at":"2026-08-08T20:08:57.256985Z","title":"Gpt-4 technical report","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":3,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.256985Z"},"links":{"cited_paper":"/paper/2303.08774","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:fb16f9b35e5b2e211318c12bfd5012ebce06ed34ddb5690a2e513f8c3bdf5725","observation_id":"b8959d6a-46b3-49eb-b5cb-c0a748f279a5","resolution":{"observed_at":"2026-08-08T20:08:57.256985Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2502.02737","last_updated":"2025-02-04T21:43:16Z","snapshot_observed_at":"2026-08-14T05:03:05.661970Z","submitted_at":"2025-02-04T21:43:16Z","title":"SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2502.02737","snapshot_observed_at":"2026-08-08T20:08:57.261674Z","title":"Smollm2: When smol goes big--data-centric training of a small language model","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":4,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.261674Z"},"links":{"cited_paper":"/paper/2502.02737","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:bb786001182a4f6a7a7f36be16d75be3760aa7e8f795c76651ae406da2eaeb40","observation_id":"8f656a0f-a84b-413c-b257-814d9beb584a","resolution":{"observed_at":"2026-08-08T20:08:57.261674Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.218542Z","title":"Physics of language models: Part 3.3, knowledge capacity scaling laws","venue":null,"work_id":"7eaddd13-e74f-456e-84ab-edeb6a28442d","year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.265845Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:75f1ebd5af371a122afd8842fff9016f630125ee4105026dd2c0f00b44b0ff7e","observation_id":"4dd82cc8-3118-4921-b6f6-d25a72487b88","resolution":{"observed_at":"2026-08-08T20:08:58.222638Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2108.01624","last_updated":"2021-08-03T16:51:36Z","snapshot_observed_at":"2026-08-13T18:34:35.124329Z","submitted_at":"2021-08-03T16:51:36Z","title":"Large-Scale Differentially Private BERT","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2108.01624","snapshot_observed_at":"2026-08-08T20:08:57.270395Z","title":"Large-scale differentially private bert","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":6,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.270395Z"},"links":{"cited_paper":"/paper/2108.01624","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:478c1187d1d44d59aef2fcf4c3f30642da1528e9806ea30af8356fba0eb923c4","observation_id":"807bf33c-57d7-452b-9f79-7ea34005f6ae","resolution":{"observed_at":"2026-08-08T20:08:57.270395Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2402.09327","last_updated":"2024-07-18T17:37:59Z","snapshot_observed_at":"2026-08-13T04:18:56.331547Z","submitted_at":"2024-02-14T17:17:30Z","title":"Information Complexity of Stochastic Convex Optimization: Applications to Generalization and Memorization","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2402.09327","snapshot_observed_at":"2026-08-08T20:08:57.275663Z","title":"Information complexity of stochastic convex optimization: Applications to generalization and memorization","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":7,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.275663Z"},"links":{"cited_paper":"/paper/2402.09327","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:45ae800fed1b38ea2870dc13dafdfa018c52368fdb05be6b5ca30fcea305ea78","observation_id":"89d0b8c9-37f5-4fad-afc5-bfc94d30165e","resolution":{"observed_at":"2026-08-08T20:08:57.275663Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2007.14966","last_updated":"2021-01-14T21:36:02Z","snapshot_observed_at":"2026-08-09T05:16:30.145678Z","submitted_at":"2020-07-29T17:22:26Z","title":"Mirostat: A Neural Text Decoding Algorithm that Directly Controls Perplexity","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2007.14966","snapshot_observed_at":"2026-08-08T20:08:57.279687Z","title":"Mirostat: A neural text decoding algorithm that directly controls perplexity","venue":null,"work_id":null,"year":2007},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":8,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.279687Z"},"links":{"cited_paper":"/paper/2007.14966","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:10bf91039eed745f73dcd1ac58a7e507b4eb1256670cab6c51876d1cb52fcc64","observation_id":"349d2339-b466-4c0b-aaee-38d5a8e5f8ca","resolution":{"observed_at":"2026-08-08T20:08:57.279687Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2304.01373","last_updated":"2023-05-31T17:54:07Z","snapshot_observed_at":"2026-08-14T19:44:03.879127Z","submitted_at":"2023-04-03T20:58:15Z","title":"Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2304.01373","snapshot_observed_at":"2026-08-08T20:08:57.283516Z","title":"Pythia: A suite for analyzing large language models across training and scaling, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":9,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.283516Z"},"links":{"cited_paper":"/paper/2304.01373","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:23fce37649a60696cf71fccbdf42db0ceac5d55ffdedab502423627ac832e4b2","observation_id":"195b94fb-221a-4670-b05a-4515bac6a188","resolution":{"observed_at":"2026-08-08T20:08:57.283516Z","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":"2026-08-08T20:08:57.287563Z","title":"Emergent and predictable memorization in large language models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":10,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.287563Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:49dc1cafba088af728b713763d5a4f819abef8a99a8ad309c246028924416d68","observation_id":"aa61a99b-9f3a-420e-961d-f6a993ed8ae5","resolution":{"observed_at":"2026-08-08T20:08:57.287563Z","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":"2026-08-08T20:08:57.291266Z","title":"Piqa: Reasoning about physical commonsense in natural language","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":11,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.291266Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:140e2b2a80e045f3b0136a4c4f27c6b7e7875511ca6e003e594282f5d7ece4d1","observation_id":"34aadf66-5f4e-41c8-bac9-13250cbdf6f1","resolution":{"observed_at":"2026-08-08T20:08:57.291266Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.195619Z","title":"Wikipedia, the free encyclopedia","venue":null,"work_id":"767f576b-7e9f-4ab2-be1a-e0a6ab58ab10","year":2001},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":12,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.295017Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:3e774e599b6065700607777e73eb1459198d4bfd854375be6949a304f1fe4eb8","observation_id":"47d69e1c-7096-48a4-ad8c-3cb3c15ea156","resolution":{"observed_at":"2026-08-08T20:08:58.200038Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.185329Z","title":"Targeted memorized‐data unlearning for large language models","venue":null,"work_id":"2fa7b7ca-6225-4115-bbe4-4edb626588e6","year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":13,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.298418Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:c3d02a0a290e89c0d3106f03262a1477a6be3eb552b15266b26b4f34557b91f2","observation_id":"fa722a2d-d573-47f2-87ee-29a7f5ee5fb1","resolution":{"observed_at":"2026-08-08T20:08:58.189060Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:57.301920Z","title":"Extracting training data from large language models","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":14,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.301920Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:a4d47917e29f7c1af8ffa525003f005e77a7c8a1511ea1b6a83109972ef1a823","observation_id":"eb2ef0b0-5d0f-4317-a106-b4164bd961e0","resolution":{"observed_at":"2026-08-08T20:08:57.301920Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2202.07646","last_updated":"2023-03-06T06:28:18Z","snapshot_observed_at":"2026-08-03T19:51:05.627063Z","submitted_at":"2022-02-15T18:48:31Z","title":"Quantifying Memorization Across Neural Language Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2202.07646","snapshot_observed_at":"2026-08-08T20:08:57.305404Z","title":"Quantifying memorization across neural language models","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":15,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.305404Z"},"links":{"cited_paper":"/paper/2202.07646","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:0721cb714f8fca91ccfbd2e9be291f6174ba08ae1b0c3016fd18e73f7ce87962","observation_id":"8fe3a03b-f428-4690-93e2-1dce58bb4ce5","resolution":{"observed_at":"2026-08-08T20:08:57.305404Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.167722Z","title":"Do localization methods actually localize memorized data in llms? a tale of two benchmarks","venue":null,"work_id":"0caa91aa-e60a-48ed-a163-90b184f8c4f3","year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":16,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.309496Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:14daa598e5787e7775bde51a244200534e27a9b9fb4d7bcebf1a73581868cca2","observation_id":"09fe5233-3b88-484a-a2e0-750c08b3e6a0","resolution":{"observed_at":"2026-08-08T20:08:58.172298Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.157221Z","title":"Neural surgery for memorisation: Locating and removing verbatim recall neurons","venue":null,"work_id":"24b2abb6-df40-41df-a17e-344aea3fa2ba","year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":17,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.313147Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:6b5208808e0185494d9ca9aedb1c5c2bd3bb7f9ab2b6d10d95010c42a11c848b","observation_id":"a3efbc0b-3346-43bc-9782-15ed867e906e","resolution":{"observed_at":"2026-08-08T20:08:58.161104Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2308.08066","last_updated":"2023-08-15T22:49:30Z","snapshot_observed_at":"2026-08-15T00:53:39.633537Z","submitted_at":"2023-08-15T22:49:30Z","title":"The Geometry of Constant Function Market Makers","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2308.08066","snapshot_observed_at":"2026-08-08T20:08:57.318096Z","title":"Accelerating large language model decoding with speculative sampling","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":18,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.318096Z"},"links":{"cited_paper":"/paper/2308.08066","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:8a1ada0a702ff8837bb95643e02fd4f2f5b8999f4f8e587eea556a067648d354","observation_id":"65a35925-5fc3-42b2-a5d6-d2abfdab0984","resolution":{"observed_at":"2026-08-08T20:08:57.318096Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2504.14452","last_updated":"2025-07-17T18:57:22Z","snapshot_observed_at":"2026-08-09T20:05:11.909475Z","submitted_at":"2025-04-20T01:59:46Z","title":"ParaPO: Aligning Language Models to Reduce Verbatim Reproduction of Pre-training Data","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2504.14452","snapshot_observed_at":"2026-08-08T20:08:57.322125Z","title":"Parapo: Aligning language models to reduce verbatim reproduction of pre-training data","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":19,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.322125Z"},"links":{"cited_paper":"/paper/2504.14452","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:f211671134eaa96121cd5e6c7b901a06cf0340acd5b140c8fe1f409bf61d22ee","observation_id":"3142da95-84b7-45da-9503-ac79ab796a6d","resolution":{"observed_at":"2026-08-08T20:08:57.322125Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1905.10044","last_updated":"2019-05-24T05:48:49Z","snapshot_observed_at":"2026-08-14T15:52:43.138964Z","submitted_at":"2019-05-24T05:48:49Z","title":"BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1905.10044","snapshot_observed_at":"2026-08-08T20:08:57.325819Z","title":"Boolq: Exploring the surprising difficulty of natural yes/no questions","venue":null,"work_id":null,"year":1905},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":20,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.325819Z"},"links":{"cited_paper":"/paper/1905.10044","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:7f3e51e1b6a972e0e596699afb286d9476e1c056a79a93bf929f66e5e556def7","observation_id":"2b70624b-e620-4e46-8066-e6a7ed0db8c3","resolution":{"observed_at":"2026-08-08T20:08:57.325819Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1803.05457","last_updated":"2018-03-14T18:04:21Z","snapshot_observed_at":"2026-08-14T19:36:07.505691Z","submitted_at":"2018-03-14T18:04:21Z","title":"Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1803.05457","snapshot_observed_at":"2026-08-08T20:08:57.330593Z","title":"Think you have solved question answering? try arc, the ai2 reasoning challenge","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":21,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.330593Z"},"links":{"cited_paper":"/paper/1803.05457","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:7c791fa3d73a719674ed4c071194ffa9f4d82d80e9a7e461ac5ed316ce4f5976","observation_id":"4eb2ecef-7cf6-4752-92a2-f12810a78d74","resolution":{"observed_at":"2026-08-08T20:08:57.330593Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.146262Z","title":"The corpus of contemporary american english as the first reliable monitor corpus of english","venue":null,"work_id":"f1c2cfe6-7ca8-42f9-8f8e-86b5f56afa30","year":2010},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":22,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.334536Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:9b8e5c678ee2a36aa8c6170a1f9afb241c1470eed952c6003616f15b5adad72c","observation_id":"cb1b3609-a6cc-40e8-bfae-d36ae43bf38b","resolution":{"observed_at":"2026-08-08T20:08:58.149771Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2407.21783","last_updated":"2024-11-23T23:27:33Z","snapshot_observed_at":"2026-08-13T17:20:44.002518Z","submitted_at":"2024-07-31T17:54:27Z","title":"The Llama 3 Herd of Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2407.21783","snapshot_observed_at":"2026-08-08T20:08:57.338218Z","title":"The llama 3 herd of models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":23,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.338218Z"},"links":{"cited_paper":"/paper/2407.21783","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:e18b0df76b1a8eceeffac75d1508eabfcc798f924b601645a5ed32a12e3876fe","observation_id":"00fc16e0-f1da-4091-8a71-99ac84cfb233","resolution":{"observed_at":"2026-08-08T20:08:57.338218Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.02238","last_updated":"2023-10-04T05:20:19Z","snapshot_observed_at":"2026-08-13T05:58:13.172138Z","submitted_at":"2023-10-03T17:48:14Z","title":"Who's Harry Potter? Approximate Unlearning in LLMs","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.02238","snapshot_observed_at":"2026-08-08T20:08:57.341933Z","title":"Who’s harry potter? approximate unlearning in llms","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":24,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.341933Z"},"links":{"cited_paper":"/paper/2310.02238","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:675a0b9384c72c3b95d14f0a3c5660ad88107d969623f15fbc60d5fd72987151","observation_id":"eb44ab29-1486-44c8-b049-ea1d1b5aed00","resolution":{"observed_at":"2026-08-08T20:08:57.341933Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2305.14822","last_updated":"2024-03-24T13:01:18Z","snapshot_observed_at":"2026-08-13T11:34:48.488876Z","submitted_at":"2023-05-24T07:22:41Z","title":"Can Copyright be Reduced to Privacy?","version":2},"cited_work":{"arxiv_id":"2305.14822","doi":null,"metadata_source":"pith","pith_arxiv_id":"2305.14822","snapshot_observed_at":"2026-08-08T20:08:57.810468Z","title":"Can Copyright be Reduced to Privacy?","venue":"cs.LG","work_id":"c556b21c-45fb-4820-b8c0-e33d1583ac2c","year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":25,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.345598Z"},"links":{"cited_paper":"/paper/2305.14822","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:3aa316b8fe6ef272d8aab0a8162c139e332eb1b55e7be25151e6512aa9be0e4f","observation_id":"acac0fc0-d61c-4747-a5ea-b52e3da919a6","resolution":{"observed_at":"2026-08-08T20:08:57.814487Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1805.04833","last_updated":"2018-05-13T07:07:08Z","snapshot_observed_at":"2026-08-14T19:15:59.926210Z","submitted_at":"2018-05-13T07:07:08Z","title":"Hierarchical Neural Story Generation","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1805.04833","snapshot_observed_at":"2026-08-08T20:08:57.350286Z","title":"Hierarchical neural story generation","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":26,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.350286Z"},"links":{"cited_paper":"/paper/1805.04833","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:5c30abb2fab4a4dea47527bb5eb8a13efe99906d7de7b642e72e74c0471647ac","observation_id":"5d7397fc-d826-4601-b4e7-d9985e5fbe0e","resolution":{"observed_at":"2026-08-08T20:08:57.350286Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2412.06370","last_updated":"2024-12-09T10:44:47Z","snapshot_observed_at":"2026-08-13T14:19:29.737181Z","submitted_at":"2024-12-09T10:44:47Z","title":"Exploring Memorization and Copyright Violation in Frontier LLMs: A Study of the New York Times v. OpenAI 2023 Lawsuit","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2412.06370","snapshot_observed_at":"2026-08-08T20:08:57.353850Z","title":"Exploring memorization and copyright violation in frontier llms: A study of the new york times v","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":27,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.353850Z"},"links":{"cited_paper":"/paper/2412.06370","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:d05d8f8939ab0f9c851d5be14e4a35b2d1d4e4641d5674b5eba5a0ebf0fec7c8","observation_id":"e5b20004-ccb1-4fc0-8109-bf49ba8a772f","resolution":{"observed_at":"2026-08-08T20:08:57.353850Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.135578Z","title":"The times sues openai and microsoft over ai use of copyrighted work","venue":null,"work_id":"0987f42c-e55f-4fa5-8cde-0444e283e8a2","year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":28,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.357649Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:2370dc9a9b73ccdb106f1bdc7d9af6b0d575defa873b94bb80275c008f53b3f5","observation_id":"25398975-271e-46e0-912b-3f11b98d8bb5","resolution":{"observed_at":"2026-08-08T20:08:58.139277Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.124852Z","title":"Leetcode problem dataset, 2021","venue":null,"work_id":"c5a33de4-f9fb-4f7b-b3ec-ffb7fcf82916","year":2021},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":29,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.361669Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:bce5fef8a00747d1789b7c2f48e3a62ca97685167346db820c5a113c91f8973e","observation_id":"daadf5f6-4e4c-40d4-87f5-1f1f1d074a43","resolution":{"observed_at":"2026-08-08T20:08:58.128737Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2406.10209","last_updated":"2024-11-02T23:19:18Z","snapshot_observed_at":"2026-08-15T02:43:18.018213Z","submitted_at":"2024-06-14T17:44:22Z","title":"Be like a Goldfish, Don't Memorize! Mitigating Memorization in Generative LLMs","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2406.10209","snapshot_observed_at":"2026-08-08T20:08:57.365034Z","title":"Be like a goldfish, don't memorize! mitigating memorization in generative llms","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":30,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.365034Z"},"links":{"cited_paper":"/paper/2406.10209","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:a5881af2419d8a0f134f8eabd96091914d35ca34817adb041c05c8095690c1b7","observation_id":"3d26e1d3-e9c8-4333-b763-5ee1d05470bd","resolution":{"observed_at":"2026-08-08T20:08:57.365034Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.18362","last_updated":"2023-10-24T14:25:53Z","snapshot_observed_at":"2026-08-14T13:12:44.064754Z","submitted_at":"2023-10-24T14:25:53Z","title":"SoK: Memorization in General-Purpose Large Language Models","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.18362","snapshot_observed_at":"2026-08-08T20:08:57.369870Z","title":"Sok: Memorization in general-purpose large language models, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":31,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.369870Z"},"links":{"cited_paper":"/paper/2310.18362","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:8906490f478e01e91ad1e959d1a0bbd4a2d07fda65f8dd9e678f3a63675a7c9b","observation_id":"0b5f86ce-1c4d-4fd1-aef5-8d253e9d7ad4","resolution":{"observed_at":"2026-08-08T20:08:57.369870Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2304.01933","last_updated":"2023-10-09T15:38:46Z","snapshot_observed_at":"2026-08-13T12:09:20.320930Z","submitted_at":"2023-04-04T16:31:37Z","title":"LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2304.01933","snapshot_observed_at":"2026-08-08T20:08:57.373671Z","title":"Llm-adapters: An adapter family for parameter-efficient fine-tuning of large language models, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":32,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.373671Z"},"links":{"cited_paper":"/paper/2304.01933","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:3f417c72e7095d554e20a15d930de072782282f997dffd4ac05bff0bdd5b43ee","observation_id":"23a1376e-4b33-4bb5-86f9-22277cf77a6f","resolution":{"observed_at":"2026-08-08T20:08:57.373671Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2407.17817","last_updated":"2024-07-25T07:10:31Z","snapshot_observed_at":"2026-08-13T00:52:52.593110Z","submitted_at":"2024-07-25T07:10:31Z","title":"Demystifying Verbatim Memorization in Large Language Models","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2407.17817","snapshot_observed_at":"2026-08-08T20:08:57.377140Z","title":"Demystifying verbatim memorization in large language models, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":33,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.377140Z"},"links":{"cited_paper":"/paper/2407.17817","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:46c12433ec574652d0a710fa3002f0c819c2e6c4f0c1a104510866d91d3c635c","observation_id":"84ff37c5-46d0-4732-a194-298cbc6d3c49","resolution":{"observed_at":"2026-08-08T20:08:57.377140Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2210.17546","last_updated":"2023-09-11T16:58:48Z","snapshot_observed_at":"2026-08-13T13:52:46.042889Z","submitted_at":"2022-10-31T17:57:55Z","title":"Preventing Verbatim Memorization in Language Models Gives a False Sense of Privacy","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2210.17546","snapshot_observed_at":"2026-08-08T20:08:57.381019Z","title":"Preventing verbatim memorization in language models gives a false sense of privacy","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":34,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.381019Z"},"links":{"cited_paper":"/paper/2210.17546","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:bbded7baf7bd54c8f8bbbfb1578ce5ebdaef33dd197b44ec1d04e7c8af21f139","observation_id":"2bc12b95-f8ac-44cc-97a5-a9bb4ea893db","resolution":{"observed_at":"2026-08-08T20:08:57.381019Z","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":"2026-08-13T14:13:28.459055Z","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":"2026-08-08T20:08:57.384769Z","title":"Knowledge unlearning for mitigating privacy risks in language models","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":35,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.384769Z"},"links":{"cited_paper":"/paper/2210.01504","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:917ca333f88a31514875abf50f9f5b8933569410552823aeb4eb398fa848dfff","observation_id":"ebc86268-dd95-4bda-9aab-b04964ea159d","resolution":{"observed_at":"2026-08-08T20:08:57.384769Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.112615Z","title":"Deduplicating training data mitigates privacy risks in language models","venue":null,"work_id":"74580d96-145a-41bf-8c5b-9c436d307068","year":2022},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":36,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.389616Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:d0edf91047d6c4647dcdcba243077e2c2169600ef0a10934d3d512cb5d2c946b","observation_id":"170397e8-c715-4f70-ba57-81fbb3235fc2","resolution":{"observed_at":"2026-08-08T20:08:58.117084Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2310.13771","last_updated":"2023-10-20T19:14:59Z","snapshot_observed_at":"2026-08-14T00:47:15.826102Z","submitted_at":"2023-10-20T19:14:59Z","title":"Copyright Violations and Large Language Models","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.13771","snapshot_observed_at":"2026-08-08T20:08:57.393267Z","title":"Copyright violations and large language models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":37,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.393267Z"},"links":{"cited_paper":"/paper/2310.13771","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:be11a8e9f7148d240272e2ad6720b987336fccdbf2f7abbbcac6e58d7b789d4b","observation_id":"2f45330f-b883-462a-8846-a0c872513e62","resolution":{"observed_at":"2026-08-08T20:08:57.393267Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.100326Z","title":"Big-little decoder: Faster language generation with an auxiliary model","venue":null,"work_id":"b62ca654-bf0b-488f-b6cb-89e08d95403c","year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":38,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.396946Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:b49532b48683ce2a2f6c605faa3355cfe42a62bb20ce95397cd394ee5e0b35d6","observation_id":"665d368c-3858-4c8c-8e26-762581105b80","resolution":{"observed_at":"2026-08-08T20:08:58.105232Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.089340Z","title":"Fast inference from transformers via speculative decoding","venue":null,"work_id":"3a7cb2c9-a81a-4ca1-ac5a-3cc4911ea056","year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":39,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.401684Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:27273b0d77d98158dc7f94df692034a962059e1a6a354b1de13fb8f015eb1625","observation_id":"df7f59d6-8712-41a5-a7c4-2f16c88bcb9d","resolution":{"observed_at":"2026-08-08T20:08:58.093558Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.077939Z","title":"Contrastive decoding: Open-ended text generation as conditional density estimation","venue":null,"work_id":"5b63334d-a901-4137-91d0-30ba87b6d211","year":2022},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":40,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.405452Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:66ac7f9a42b9c17570b84d4e515ae52497ee2216fed1c659c244b3d9ec93e091","observation_id":"f90f4c37-637b-4061-80f9-90a6943510d2","resolution":{"observed_at":"2026-08-08T20:08:58.081975Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2412.19437","last_updated":"2025-02-18T17:26:38Z","snapshot_observed_at":"2026-08-11T01:48:59.557045Z","submitted_at":"2024-12-27T04:03:16Z","title":"DeepSeek-V3 Technical Report","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2412.19437","snapshot_observed_at":"2026-08-08T20:08:57.409176Z","title":"Deepseek-v3 technical report","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":41,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.409176Z"},"links":{"cited_paper":"/paper/2412.19437","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:8377b17bee012036403d985789bb4af74903fda1f6ae2f60814d2b772fb5dde5","observation_id":"7d5c5bf9-f9db-43e4-9c49-9eaaaac4dd13","resolution":{"observed_at":"2026-08-08T20:08:57.409176Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"cs/0205028","last_updated":"2002-05-17T12:51:00Z","snapshot_observed_at":"2026-07-07T03:04:36.695497Z","submitted_at":"2002-05-17T12:51:00Z","title":"NLTK: The Natural Language Toolkit","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"cs/0205028","snapshot_observed_at":"2026-08-08T20:08:57.413302Z","title":"Nltk: The natural language toolkit","venue":null,"work_id":null,"year":2002},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":42,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.413302Z"},"links":{"cited_paper":"/paper/cs/0205028","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:8554e5bbb0addb7fd01b4db0eb972599f5965afd433291b8c5c9464a4a5717a4","observation_id":"b5bc4549-1fc4-48d5-9240-7018971984d9","resolution":{"observed_at":"2026-08-08T20:08:57.413302Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2307.09542","last_updated":"2023-07-18T18:36:29Z","snapshot_observed_at":"2026-08-13T14:04:40.548339Z","submitted_at":"2023-07-18T18:36:29Z","title":"Can Neural Network Memorization Be Localized?","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2307.09542","snapshot_observed_at":"2026-08-08T20:08:57.417863Z","title":"Can neural network memorization be localized? arXiv preprint arXiv:2307.09542, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":43,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.417863Z"},"links":{"cited_paper":"/paper/2307.09542","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:be8bd30928345937183e944a9dc54eb10d98ae5d27887e144548495742db6a5e","observation_id":"ff8e80df-79af-449c-9340-e2601a2b0dd3","resolution":{"observed_at":"2026-08-08T20:08:57.417863Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1809.02789","last_updated":"2018-09-08T11:47:16Z","snapshot_observed_at":"2026-08-14T07:00:02.529732Z","submitted_at":"2018-09-08T11:47:16Z","title":"Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1809.02789","snapshot_observed_at":"2026-08-08T20:08:57.422630Z","title":"Can a suit of armor conduct electricity? a new dataset for open book question answering","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":44,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.422630Z"},"links":{"cited_paper":"/paper/1809.02789","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:265795b5582378db26f293af87308322e978a76b37bc3cfabafa1eaa1bb0fd13","observation_id":"453f33d7-069c-401b-921b-9550b708f79e","resolution":{"observed_at":"2026-08-08T20:08:57.422630Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2205.12506","last_updated":"2022-11-04T03:13:41Z","snapshot_observed_at":"2026-08-13T15:37:34.650664Z","submitted_at":"2022-05-25T05:49:31Z","title":"Memorization in NLP Fine-tuning Methods","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2205.12506","snapshot_observed_at":"2026-08-08T20:08:57.426733Z","title":"Memorization in nlp fine-tuning methods","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":45,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.426733Z"},"links":{"cited_paper":"/paper/2205.12506","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:b8eda3ddca79b88a9a80c1b53763b8185a555cf079a248400e285886c485134a","observation_id":"4be2acd5-1f7b-41da-b179-f7aca784429b","resolution":{"observed_at":"2026-08-08T20:08:57.426733Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.17035","last_updated":"2023-11-28T18:47:03Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2023-11-28T18:47:03Z","title":"Scalable Extraction of Training Data from (Production) Language Models","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.17035","snapshot_observed_at":"2026-08-08T20:08:57.430369Z","title":"Scalable extraction of training data from (production) language models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":46,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.430369Z"},"links":{"cited_paper":"/paper/2311.17035","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:85befd9bfeb21941679f2fe1fd8592f2fcb4f703ea0231bc6ca75cfa3fc235ee","observation_id":"4f427f2f-7094-4895-af91-af8225b2175a","resolution":{"observed_at":"2026-08-08T20:08:57.430369Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.066458Z","title":"Scalable extraction of training data from aligned, production language models","venue":null,"work_id":"b786f03c-3056-4702-a8d3-c949492d6fe0","year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":47,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.434194Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:f85d12542f3ccec92697d8bff4aa1edf28d48b5cfc8b212bf2731f1205fec78b","observation_id":"1ed45db8-3804-4be8-be55-6e093a165be9","resolution":{"observed_at":"2026-08-08T20:08:58.070573Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.054676Z","title":"Generative ai and copyright issues globally: Ani media v openai","venue":null,"work_id":"4425604b-1381-4e12-937a-6bfd8cdf6c72","year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":48,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.438835Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:1c10ee85d5fdf9c4ac1b21b93eb606de34f15c32c5259c276afbba982e1dc905","observation_id":"47932ff0-966d-4c0d-815d-68511430a0aa","resolution":{"observed_at":"2026-08-08T20:08:58.059329Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2410.11985","last_updated":"2024-10-15T18:47:12Z","snapshot_observed_at":"2026-08-14T14:48:58.147534Z","submitted_at":"2024-10-15T18:47:12Z","title":"The Fair Language Model Paradox","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2410.11985","snapshot_observed_at":"2026-08-08T20:08:57.442425Z","title":"The fair language model paradox","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":49,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.442425Z"},"links":{"cited_paper":"/paper/2410.11985","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:73fe3803f68f9c523c666e4ee94170a807f891852c51bdfb89aebe5fb14d0e82","observation_id":"70fa3916-a3f1-40c6-a62e-f33dbdbb5e3c","resolution":{"observed_at":"2026-08-08T20:08:57.442425Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2407.08707","last_updated":"2024-07-11T17:44:41Z","snapshot_observed_at":"2026-08-12T23:24:00.078338Z","submitted_at":"2024-07-11T17:44:41Z","title":"Extracting Training Data from Document-Based VQA Models","version":1},"cited_work":{"arxiv_id":"2407.08707","doi":null,"metadata_source":"pith","pith_arxiv_id":"2407.08707","snapshot_observed_at":"2026-08-08T20:08:57.645956Z","title":"Extracting Training Data from Document-Based VQA Models","venue":"cs.CV","work_id":"86ad6461-99df-427f-b2d2-6d6337ce0eef","year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":50,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.446287Z"},"links":{"cited_paper":"/paper/2407.08707","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:31ab3e5f991b560d5bb6c060647acc75f66e46ec98e2f0c2ad951ad64d506b1b","observation_id":"edd3352d-1ed9-4f08-a78f-fb7c65921c47","resolution":{"observed_at":"2026-08-08T20:08:57.650604Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2502.15010","last_updated":"2025-06-12T10:12:38Z","snapshot_observed_at":"2026-08-08T00:48:28.063583Z","submitted_at":"2025-02-20T20:02:56Z","title":"Obliviate: Efficient Unmemorization for Protecting Intellectual Property in Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2502.15010","snapshot_observed_at":"2026-08-08T20:08:57.451004Z","title":"Obliviate: Efficient unmemorization for protecting intellectual property in large language models","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":51,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.451004Z"},"links":{"cited_paper":"/paper/2502.15010","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:63263366ecc2e5dbf116acea9fae428f0a0c184512afccbff946477bd131874b","observation_id":"a6c2f15d-1476-443a-90d6-5cd8eb1b4817","resolution":{"observed_at":"2026-08-08T20:08:57.451004Z","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":"2026-08-08T20:08:57.454704Z","title":"Winogrande: An adversarial winograd schema challenge at scale","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":52,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.454704Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:393db8a0a69a535915133c28bcdfa8dbfa19b53c3a86331203e14ac194dedfee","observation_id":"f079b7cc-6af9-4de8-bab4-67ab6fef6fd7","resolution":{"observed_at":"2026-08-08T20:08:57.454704Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2410.02159","last_updated":"2025-01-28T21:19:25Z","snapshot_observed_at":"2026-08-12T22:31:59.473224Z","submitted_at":"2024-10-03T02:53:51Z","title":"Mitigating Memorization In Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2410.02159","snapshot_observed_at":"2026-08-08T20:08:57.457970Z","title":"Mitigating memorization in language models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":53,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.457970Z"},"links":{"cited_paper":"/paper/2410.02159","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:056bada7149261152d78953ed85d236f1dd29ff131474cf213ff4a35c781ff4b","observation_id":"b9c2bfa4-f079-4939-8372-0a2d0c15cf8d","resolution":{"observed_at":"2026-08-08T20:08:57.457970Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.037586Z","title":"Distilbert, a distilled version of bert: smaller, faster, cheaper and lighter","venue":null,"work_id":"4c738a55-1d4f-4b6d-9c85-51d5f7da44d6","year":2019},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":54,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.461956Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:4fc45f2d4fe0f26c2e5e783718e8f9acdb68496812334874c7dbe22e833b5025","observation_id":"2b8bbf0f-0763-46cd-afca-7141ce5e7945","resolution":{"observed_at":"2026-08-08T20:08:58.041547Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1904.09728","last_updated":"2019-09-09T17:29:55Z","snapshot_observed_at":"2026-08-12T21:45:41.485479Z","submitted_at":"2019-04-22T05:36:37Z","title":"SocialIQA: Commonsense Reasoning about Social Interactions","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1904.09728","snapshot_observed_at":"2026-08-08T20:08:57.465476Z","title":"Socialiqa: Commonsense reasoning about social interactions","venue":null,"work_id":null,"year":1904},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":55,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.465476Z"},"links":{"cited_paper":"/paper/1904.09728","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:6dd5add1c9130a0f913715354131f9941aa8cc8f9a64ffed853128491a98ce1d","observation_id":"63e7a741-a1e4-444b-8c21-1ff00b81d8fe","resolution":{"observed_at":"2026-08-08T20:08:57.465476Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2404.15146","last_updated":"2024-11-11T19:47:16Z","snapshot_observed_at":"2026-08-13T15:57:02.054302Z","submitted_at":"2024-04-23T15:49:37Z","title":"Rethinking LLM Memorization through the Lens of Adversarial Compression","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2404.15146","snapshot_observed_at":"2026-08-08T20:08:57.470379Z","title":"Rethinking llm memorization through the lens of adversarial compression","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":56,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.470379Z"},"links":{"cited_paper":"/paper/2404.15146","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:8d5092ab9909b5082528deb9c0b5377c94a60d874bcc229f066d2aa8f8cc7ed8","observation_id":"565692a6-1b0d-4ee9-bd2d-923e9b528af8","resolution":{"observed_at":"2026-08-08T20:08:57.470379Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2407.00106","last_updated":"2024-06-27T10:24:35Z","snapshot_observed_at":"2026-08-13T16:16:59.808911Z","submitted_at":"2024-06-27T10:24:35Z","title":"UnUnlearning: Unlearning is not sufficient for content regulation in advanced generative AI","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2407.00106","snapshot_observed_at":"2026-08-08T20:08:57.474018Z","title":"Ununlearning: Unlearning is not sufficient for content regulation in advanced generative ai","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":57,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.474018Z"},"links":{"cited_paper":"/paper/2407.00106","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:3a892493e31d8636a05c844dbf14c318ba5d6cfbff5b8390e7d775a0fa7a4162","observation_id":"8feae682-7ffa-4ad0-8276-0a9d528e4b60","resolution":{"observed_at":"2026-08-08T20:08:57.474018Z","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":"2026-08-08T20:08:57.477897Z","title":"Slimpajama: A 627b token cleaned and deduplicated version of redpajama, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":58,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.477897Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:a9d7586277b49ed4302fa016ca233ca66f595882f08a0e1ceb40c4f0840264b0","observation_id":"623241d4-a7c2-4651-af4e-a0236dcfee9f","resolution":{"observed_at":"2026-08-08T20:08:57.477897Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.018406Z","title":"Blockwise parallel decoding for deep autoregressive models","venue":null,"work_id":"443a5259-e698-4f91-95dc-0c5a0149c2be","year":2018},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":59,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.482077Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:dbcb491a287dfc6202119883e41adcd3dcea8da67bf5a6fe58cddd6acd0f4a4e","observation_id":"15894f9f-4382-49ce-8512-1eb64c032142","resolution":{"observed_at":"2026-08-08T20:08:58.022757Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:58.005148Z","title":"Activation steering: Mitigating verbatim memorisation at inference time","venue":null,"work_id":"5c67bbc2-a565-450d-b4e2-d27df65cdfb9","year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":60,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.485474Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:83192d10a1316e4ad23164b865a667c62af5b18c5fa707ecd343358bc00b3fb0","observation_id":"5c292de2-8bb3-43b4-a04f-2b1cca06b852","resolution":{"observed_at":"2026-08-08T20:08:58.010704Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2312.11805","last_updated":"2025-05-09T21:04:06Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2023-12-19T02:39:27Z","title":"Gemini: A Family of Highly Capable Multimodal Models","version":5},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2312.11805","snapshot_observed_at":"2026-08-08T20:08:57.488782Z","title":"Gemini: a family of highly capable multimodal models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":61,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.488782Z"},"links":{"cited_paper":"/paper/2312.11805","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:8e040a9ed0154ec3d27739c52101aefef5a5a4af840775345cf531d6d31e6bab","observation_id":"f4d851db-b509-484f-bc65-8f4584fcbddc","resolution":{"observed_at":"2026-08-08T20:08:57.488782Z","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":"2026-08-08T20:08:57.492847Z","title":"Memorization without overfitting: Analyzing the training dynamics of large language models","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":62,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.492847Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:44b30a581a049048b916afcf37699d83dc6d0a50e2a2bc1aea6b122de3567cee","observation_id":"07436a8c-e043-4a41-8d69-b03ddd45629d","resolution":{"observed_at":"2026-08-08T20:08:57.492847Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2504.02193","last_updated":"2025-07-28T00:57:26Z","snapshot_observed_at":"2026-08-07T16:13:16.898106Z","submitted_at":"2025-04-03T00:36:40Z","title":"More is Less: The Pitfalls of Multi-Model Synthetic Preference Data in DPO Safety Alignment","version":3},"cited_work":{"arxiv_id":"2504.02193","doi":null,"metadata_source":"pith","pith_arxiv_id":"2504.02193","snapshot_observed_at":"2026-08-08T20:08:57.572211Z","title":"More is Less: The Pitfalls of Multi-Model Synthetic Preference Data in DPO Safety Alignment","venue":"cs.AI","work_id":"2930a518-108d-4c6d-b78d-ff2253fb431f","year":2025},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":63,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.496281Z"},"links":{"cited_paper":"/paper/2504.02193","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:afb048068aad11c43a86ccaeff74c84db245e903c9c7f9e8aedaf3dfd7910873","observation_id":"26fa64ab-eb27-4147-be49-ab9e4ec20298","resolution":{"observed_at":"2026-08-08T20:08:57.577008Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2411.12372","last_updated":"2024-11-19T09:35:28Z","snapshot_observed_at":"2026-08-14T07:55:53.745138Z","submitted_at":"2024-11-19T09:35:28Z","title":"RedPajama: an Open Dataset for Training Large Language Models","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2411.12372","snapshot_observed_at":"2026-08-08T20:08:57.500140Z","title":"Redpajama: an open dataset for training large language models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":64,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.500140Z"},"links":{"cited_paper":"/paper/2411.12372","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:83137c243e4cddca12eacb48003f1b289851c52ce2a243ae7334b2c932b2a7b7","observation_id":"935cca9e-f840-4481-bfcd-ea0a1f113379","resolution":{"observed_at":"2026-08-08T20:08:57.500140Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:57.984334Z","title":"Speculative decoding for non–autoregressive neural machine translation","venue":null,"work_id":"03b8c0aa-0975-4f5b-a06a-8a72129c2a5f","year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":65,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.503958Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:fa25d9ba9e618a9378b0c4dc446e91868c7b1cb63b38d662ea82ebf0e9d0ff00","observation_id":"9330b9f7-81af-4a33-863f-0bf068b527df","resolution":{"observed_at":"2026-08-08T20:08:57.989656Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:57.969219Z","title":"Autonomous data selection with language models for mathematical texts","venue":null,"work_id":"ba79b7b1-58fe-4575-be50-10b1856965b3","year":2024},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":66,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.507808Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:2d6b6c1e278677667286d0fd7f23bc12fe287122011f3324a80dae60132ae043","observation_id":"6c4494c2-3c03-4a81-b066-03fbe3dc81ad","resolution":{"observed_at":"2026-08-08T20:08:57.974016Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-08T20:08:57.511375Z","title":"Judging llm-as-a-judge with mt-bench and chatbot arena","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":67,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.511375Z"},"links":{"citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:777a42978dd59172c196eb8d9e7809c3d35565be5f92e3adfeea75b26a2e1702","observation_id":"d0a22b50-317f-4ca9-880d-f5a3ad805715","resolution":{"observed_at":"2026-08-08T20:08:57.511375Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2308.15727","last_updated":"2023-11-05T13:44:55Z","snapshot_observed_at":"2026-08-13T10:24:32.506731Z","submitted_at":"2023-08-30T03:06:47Z","title":"Quantifying and Analyzing Entity-level Memorization in Large Language Models","version":2},"cited_work":{"arxiv_id":"2308.15727","doi":null,"metadata_source":"pith","pith_arxiv_id":"2308.15727","snapshot_observed_at":"2026-08-08T20:08:57.544161Z","title":"Quantifying and Analyzing Entity-level Memorization in Large Language Models","venue":"cs.CL","work_id":"3ea93a24-650b-407d-9d56-5c21a93b10c9","year":2023},"citing_paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM","version":2},"reference_index":68,"source":"arxiv_source","source_observed_at":"2026-08-08T20:08:57.514936Z"},"links":{"cited_paper":"/paper/2308.15727","citing_paper":"/paper/2502.05159"},"observation_digest":"sha256:66bb48202af256d099928fceb195d8e5d2e666d93865f7eec261b89c6cb0f3b7","observation_id":"51057508-22bd-47cd-981a-bb7996a563b3","resolution":{"observed_at":"2026-08-08T20:08:57.550482Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"state":"measured"}}],"paper":{"arxiv_id":"2502.05159","last_updated":"2025-05-27T16:54:17Z","latest_version":2,"primary_category":"cs.LG","snapshot_observed_at":"2026-08-14T04:57:22.818850Z","submitted_at":"2025-02-07T18:41:21Z","title":"A Lightweight Method to Disrupt Memorized Sequences in LLM"},"reference_resolution":{"displayed":68,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":44,"verified_exact":5,"verified_fuzzy":19},"total_outbound_references":68},"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":"2026-08-14T06:32:32.682623+00:00","source":"crossref"},{"observed_at":"2026-08-14T06:32:18.44784+00:00","source":"retraction_watch"}],"thesis":"As of 15 August 2026, this Paper Citation Record lists 68 of 68 outbound references and 0 inbound Pith citation observations for arXiv:2502.05159."}