{"as_of":"2026-08-10T20:03:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:9dbfb80c906815974c11f6f02a7cf9b27e635d6fa6bfc5deeca03af5c54f82a7","coverage":[{"denominator":38,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":38,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-09T11:12:08.289178Z","state":"measured"},{"denominator":42,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":42,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-10T06:31:04.303077+00:00","state":"measured"},{"denominator":4,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":4,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-03T05:52:38.387523Z","state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"arxiv_reference","source_observed_at":"2026-05-20T07:18:07.045532Z","state":"measured"}],"external_citation_measurements":[],"inbound":[{"citation":{"cited_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2502.02797","snapshot_observed_at":"2026-08-03T05:52:38.387523Z","title":"Sanyal, S., Prairie, H., Das, R., Kavis, A., and Sanghavi, S","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2602.01058","last_updated":"2026-05-28T06:23:31Z","snapshot_observed_at":"2026-08-03T05:52:36.155371Z","submitted_at":"2026-02-01T06:53:45Z","title":"Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning","version":2},"reference_index":2020,"source":"pdf_text","source_observed_at":"2026-08-03T05:52:38.387523Z"},"links":{"cited_paper":"/paper/2502.02797","citing_paper":"/paper/2602.01058"},"observation_digest":"sha256:24aaf5fd0fe1e2c85e1d427f4eeaa8169d97506f39021f7172ab019b3153798a","observation_id":"80febf3b-4f6f-4f2d-bfb4-44f361f16f20","resolution":{"observed_at":"2026-08-03T05:52:38.387523Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"cited_work":{"arxiv_id":"2502.02797","doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":"2502.02797","snapshot_observed_at":"2026-06-05T21:23:00.469572Z","title":"Upweighting easy samples in fine-tuning mitigates forgetting","venue":null,"work_id":"b722dbea-44a8-4f8a-a59a-565f6adf4fb1","year":2025},"citing_paper":{"arxiv_id":"2604.08519","last_updated":"2026-04-09T17:55:50Z","snapshot_observed_at":"2026-07-06T22:57:35.435713Z","submitted_at":"2026-04-09T17:55:50Z","title":"Cram Less to Fit More: Training Data Pruning Improves Memorization of Facts","version":1},"reference_index":75,"source":"arxiv_source","source_observed_at":"2026-05-10T17:42:31.465077Z"},"links":{"cited_paper":"/paper/2502.02797","citing_paper":"/paper/2604.08519"},"observation_digest":"sha256:c8a34e18b993bff08ae6617a7ff1b2b3c52b09309539678a3d188206cf9ea52c","observation_id":"54d57dfe-4679-4273-a596-261ae3d82390","resolution":{"observed_at":"2026-05-11T06:15:58.986795Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"cited_work":{"arxiv_id":"2502.02797","doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":"2502.02797","snapshot_observed_at":"2026-06-05T21:23:00.469572Z","title":"Upweighting easy samples in fine-tuning mitigates forgetting","venue":null,"work_id":"b722dbea-44a8-4f8a-a59a-565f6adf4fb1","year":2025},"citing_paper":{"arxiv_id":"2605.20005","last_updated":"2026-05-19T15:36:52Z","snapshot_observed_at":"2026-07-06T23:30:39.512029Z","submitted_at":"2026-05-19T15:36:52Z","title":"Fine-Tuning Without Forgetting via Loss-Adaptive Learning Rates","version":1},"reference_index":51,"source":"pdf_text","source_observed_at":"2026-05-20T07:14:59.396900Z"},"links":{"cited_paper":"/paper/2502.02797","citing_paper":"/paper/2605.20005"},"observation_digest":"sha256:4679e4c6baa01f21bc2e2442584147eeaeb8c6f80da7c2fde63efdbedc0bf424","observation_id":"f34a04a3-6780-4f9f-9e69-7b280cef8f84","resolution":{"observed_at":"2026-05-20T07:18:07.047290Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2502.02797","snapshot_observed_at":"2026-07-31T22:25:12.430501Z","title":"Upweighting easy sam- ples in fine-tuning mitigates forgetting.arXiv preprint arXiv:2502.02797, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.24157","last_updated":"2026-07-27T08:35:09Z","snapshot_observed_at":"2026-08-10T02:10:51.662355Z","submitted_at":"2026-07-27T08:35:09Z","title":"UniGen-AR: Unifying Visual Generation with Auto-Regressive Modeling","version":1},"reference_index":57,"source":"pdf_text","source_observed_at":"2026-07-31T22:25:12.430501Z"},"links":{"cited_paper":"/paper/2502.02797","citing_paper":"/paper/2607.24157"},"observation_digest":"sha256:9c732fe0a6fcb69aa4d1e55714d3bb279b1a276712277eb04b7039cdde1b3a17","observation_id":"8e65ae95-452e-483c-bc48-f69654095d5b","resolution":{"observed_at":"2026-07-31T22:25:12.430501Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"links":{"evidence":"/evidence","html":"/paper/2502.02797/citation-record","integrity":"/paper/2502.02797/integrity","json":"/paper/2502.02797/citation-record.json","paper":"/paper/2502.02797"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-09T11:12:16.054173Z","title":"HellaSwag presents a context followed by several plausible endings, and the model must choose the most appropriate continuation","venue":null,"work_id":"25f2bae0-d7d1-4fde-90da-9d1ccd1dd0ae","year":2019},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.232279Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:3fa46e80778d4b6f6db6c5aa437afbd8980d897b41a6a09f6c857506507bdcee","observation_id":"c2ccc43e-9e2b-4e7c-925c-f5bb6aa4c785","resolution":{"observed_at":"2026-08-09T11:12:16.058577Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:16.076080Z","title":"exp − ⟨r,z⟩ 2 α ! ⟨r,z⟩ 2 # r+ d−1X j=1 E","venue":null,"work_id":"95eb3fa6-3b9e-405b-b8ea-541f2a65a94d","year":2016},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.224143Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:747e86c72d0f42a17d587afd358667e13a7efaae81f196bd2e62158dd3a17413","observation_id":"6d7fa0af-16b3-4db9-8843-7d4579950811","resolution":{"observed_at":"2026-08-09T11:12:16.080018Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:16.031989Z","title":null,"venue":null,"work_id":"c0a6126c-448c-4fb6-aff1-ecc0173c0b6f","year":2018},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.239211Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:7efea87e4276c8af17415449efab4c8c0fe9941642b751adbffa92263b7ee279","observation_id":"fc444b66-8d68-4f14-9717-5498278db48a","resolution":{"observed_at":"2026-08-09T11:12:16.035563Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2407.21783","last_updated":"2024-11-23T23:27:33Z","snapshot_observed_at":"2026-08-10T16:40:37.411115Z","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":"2026-08-09T11:12:08.165381Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.165381Z"},"links":{"cited_paper":"/paper/2407.21783","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:316df6b74964ab27f59f4cc4c68c59e484703fbe9e0c625d8c5730194b0ecea3","observation_id":"b9f055f2-d46f-46ec-bbc1-a89304aac1be","resolution":{"observed_at":"2026-08-09T11:12:08.165381Z","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":"2026-08-09T11:12:16.009800Z","title":null,"venue":null,"work_id":"f0055217-2fef-4964-b64f-7dc0bf09a5b8","year":2019},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.246306Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:55a984dea3df0fb1cf57997cd4c8de90878c19de8a032f80787785b71c8a184d","observation_id":"6f256338-d201-4106-b555-34eb22ff346a","resolution":{"observed_at":"2026-08-09T11:12:16.014017Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.999152Z","title":null,"venue":null,"work_id":"2584810b-dbff-4590-9a75-6aa40c7999ef","year":2018},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.249177Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:d36821adabafae7bba9de9aee579c6101b4d120e3aebdc7028a7d49887092b16","observation_id":"854a8fd4-5572-425b-b969-34600f2fd6d3","resolution":{"observed_at":"2026-08-09T11:12:16.002966Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.988918Z","title":null,"venue":null,"work_id":"b91af49d-7c51-4e83-9395-1c0d8e9a4520","year":null},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.252046Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:4d22aadc4fdf0a8578f1cd79ff1a68c020cbc28b62a8ebcafedbd46bd247c699","observation_id":"50348aa0-de56-4d17-8122-64802b448311","resolution":{"observed_at":"2026-08-09T11:12:15.992760Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.976482Z","title":null,"venue":null,"work_id":"5bee1e08-c290-45bb-a717-78a724e0c532","year":2021},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.255234Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:828a6ae3e8647900411e94465bd8540bb77cb65dc974aa9884b646ec29f83b62","observation_id":"ff7b63de-605d-4509-afb8-954cb436f66d","resolution":{"observed_at":"2026-08-09T11:12:15.981055Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-09T11:12:15.712387Z","title":"org/CorpusID:215786151","venue":null,"work_id":"5e8a4ba2-5488-4a82-960e-5fde7a930e6c","year":2019},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.185545Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:4d586ec72ef47c545750fc9a1963d010d74cdf9085dcf0ab433bfaf73f3b664a","observation_id":"d1137124-7b6e-47f1-9110-4e3af87fa5a3","resolution":{"observed_at":"2026-08-09T11:12:15.715919Z","resolver_source":"arxiv_id_nonexistent","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1511.06343","last_updated":"2016-04-25T14:00:21Z","snapshot_observed_at":"2026-08-10T13:42:16.073324Z","submitted_at":"2015-11-19T20:24:09Z","title":"Online Batch Selection for Faster Training of Neural Networks","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1511.06343","snapshot_observed_at":"2026-08-09T11:12:08.188932Z","title":"ISBN 9781510860964","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.188932Z"},"links":{"cited_paper":"/paper/1511.06343","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:990bebbd9a3af7eaa025acb27c6f7d046adf3807f2b94d8ef9d45169a22267e2","observation_id":"edf95260-5ff9-4d74-a7bf-98bf98702169","resolution":{"observed_at":"2026-08-09T11:12:08.188932Z","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":"biblio/1038970","doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-09T11:12:09.785656Z","title":"org/CorpusID:5324823","venue":null,"work_id":"17d3ec6e-b296-49f0-bbba-09ca95e332ee","year":2021},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.192820Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:be6ce5526ec06103283469bae9212e323754375f152e43346039a4ba2c84a705","observation_id":"bdf3674e-d94a-4e7f-ad11-9bcb54f29261","resolution":{"observed_at":"2026-08-09T11:12:09.791349Z","resolver_source":"raw_fallback","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:16.098803Z","title":"Rajasegaran, J., Hayat, M., Khan, S","venue":null,"work_id":"dc5b9cc9-3ef4-44fb-a48c-acc86db61936","year":2019},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.200188Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:b138759539f043633b4136d793e8c37c36385b4f0bb51c405b97460256cac7e8","observation_id":"412af3d0-4e8a-4aa3-90e3-163c718e1d68","resolution":{"observed_at":"2026-08-09T11:12:16.103297Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1606.04671","last_updated":"2022-10-22T14:34:44Z","snapshot_observed_at":"2026-08-09T14:14:13.613085Z","submitted_at":"2016-06-15T08:20:51Z","title":"Progressive Neural Networks","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1606.04671","snapshot_observed_at":"2026-08-09T11:12:08.204038Z","title":"org/CorpusID:245007201","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.204038Z"},"links":{"cited_paper":"/paper/1606.04671","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:052f10719c2149225e48923bf0b001d108c60a868de0dc49b2f567ee7633fdde","observation_id":"f41b576d-52a8-43f4-bedf-996de2be4cde","resolution":{"observed_at":"2026-08-09T11:12:08.204038Z","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":"2026-08-09T11:12:08.208134Z","title":"org/CorpusID:12253672","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.208134Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:bbea5455a479020c2906182c179d8b32b94ff10bc8fc5445fab3e66184c8ee37","observation_id":"771fc37b-44ac-4711-90e8-e94bb3d7cec9","resolution":{"observed_at":"2026-08-09T11:12:08.208134Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2309.12284","last_updated":"2024-05-03T17:36:07Z","snapshot_observed_at":"2026-08-02T15:00:50.388422Z","submitted_at":"2023-09-21T17:45:42Z","title":"MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Models","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2309.12284","snapshot_observed_at":"2026-08-09T11:12:08.220099Z","title":"important","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.220099Z"},"links":{"cited_paper":"/paper/2309.12284","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:4c1d771260a8973a375066b9d0cca95bbb258d3bd26932842266adffd628c601","observation_id":"99d97692-4d4d-4e3f-b9f0-c456f257e38b","resolution":{"observed_at":"2026-08-09T11:12:08.220099Z","resolver_source":null,"status":"malformed_identifier"},"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":"2026-08-09T11:12:16.065586Z","title":"Following a similar setup as us, Biderman et al","venue":null,"work_id":"d63df510-b716-430e-90bb-5f2e7c1747d9","year":2024},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.228085Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:76d231d8679bbd5b1f174f1610dd37dcf8e8361554675c454dacd157fd1d8595","observation_id":"a8f477e3-57e1-4cf2-bd67-b1c1b4467abd","resolution":{"observed_at":"2026-08-09T11:12:16.069410Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:16.042698Z","title":null,"venue":null,"work_id":"b80d343e-240b-46f1-b648-0b6c91889256","year":2018},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.235916Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:faba81a79482024e81a46483981a397b0bf9481a3946180fe47bdb25ea771e6a","observation_id":"a8474ce1-bc1b-4281-87aa-3b3011465a60","resolution":{"observed_at":"2026-08-09T11:12:16.047304Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:16.020927Z","title":"PIQA presents a goal and two possible solutions, requiring models to choose the most appropriate solution that demonstrates an understanding of everyday physical interactions","venue":null,"work_id":"dbb3c838-af88-49eb-b2c6-b926267980bd","year":2020},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.242840Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:665e1dcbd03705986b637d9d8da0ca80d072aedbc704fae4e2ed5673f59237e9","observation_id":"6daab356-2370-40f1-b67c-bdd79792b22c","resolution":{"observed_at":"2026-08-09T11:12:16.024669Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.965572Z","title":null,"venue":null,"work_id":"8546c53f-2b3d-4fa9-8587-abb7d3a1a852","year":2021},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.258173Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:2dbe4e9712713bd4cae414f8f7cd01c7b544105ee7cc0f2c531d1caa5d7c7e1d","observation_id":"0c4ec655-c716-4224-af00-562ee86007e5","resolution":{"observed_at":"2026-08-09T11:12:15.969356Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.954386Z","title":"It is a widely used large-scale image classification dataset, consisting of over a million images spanning 1000 classes","venue":null,"work_id":"5bd38c2a-a40f-4d41-a53e-21a028f4a6d1","year":2015},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.261120Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:42bd306a34a1f4cda03c97a06b716f1cb4166933fde85a069346178fa0f732f8","observation_id":"09131df9-59a1-4024-90a3-d6029de9960a","resolution":{"observed_at":"2026-08-09T11:12:15.958626Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.942155Z","title":"It consists of 60,000 32x32 color images divided into ten classes, with 6,000 images per class","venue":null,"work_id":"c26624ae-80a8-48b1-8d72-706abdeecd90","year":2009},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.264046Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:cdd96b00306111d8033354039f876c4fe86ee3984e645e4d1b097210577bb53d","observation_id":"7308244a-cde9-46bc-9e28-6d2a656b94fc","resolution":{"observed_at":"2026-08-09T11:12:15.946836Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.930077Z","title":"This dataset is used for fine-grained image classification tasks","venue":null,"work_id":"340822e4-a264-450f-bd7d-544d6f379a50","year":2009},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.266836Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:0468d2410386321b91dc98b3609a0748034465b0c0deae362332a66719a5a79d","observation_id":"b9efab78-7e0e-425c-9894-dd53841d50ee","resolution":{"observed_at":"2026-08-09T11:12:15.934177Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.919036Z","title":null,"venue":null,"work_id":"93dac327-ea82-475c-b0d7-ebc23311ae70","year":2022},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.269421Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:b8291cbf3083e3e028d07f9856b536527367a0933bd8627f77524f29fe6a1567","observation_id":"6d8d407e-5051-4bc6-b42f-ebc083f36932","resolution":{"observed_at":"2026-08-09T11:12:15.922736Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.907285Z","title":"This dataset is commonly used for fine-grained image classification and flower recognition tasks","venue":null,"work_id":"e37355ba-4d48-44bf-b2d2-c553b5c413d5","year":2008},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.272645Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:ff2110865d9bffd2f0ef8971d0d4f2db78ccf156b4e50f64dee5cbc088be84b5","observation_id":"137723b8-1704-4e20-a93c-84ab1f940b38","resolution":{"observed_at":"2026-08-09T11:12:15.911767Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.896574Z","title":"It provides a rich resource for fine-grained car classification task","venue":null,"work_id":"b46c2798-c158-43bf-99e2-1f79a3345a5c","year":2013},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.276774Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:bee3fa751735808ee31f0d5f0788ae5deed0c697e08e691458f8b7543359e117","observation_id":"cf488483-90c7-4328-a29d-3a14f6ebb672","resolution":{"observed_at":"2026-08-09T11:12:15.900306Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.884872Z","title":"This dataset is widely used for fine-grained dog breed classification and recognition tasks","venue":null,"work_id":"19f9a2b2-0276-4cee-a794-f170dc97235c","year":2012},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.281211Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:1c9a8246b91b6cd7ca6fa94783ec761fafaa39a3d40e9be21232d8ad142b8fd4","observation_id":"362cd836-dbf1-4197-9ac2-485d88f2f590","resolution":{"observed_at":"2026-08-09T11:12:15.889747Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-09T11:12:09.612291Z","title":"learning without forgetting","venue":null,"work_id":"1aa83c71-39ca-409f-8dfe-bf14a3466cb2","year":2016},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.284451Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:e9e270276481f94972001a591d0b3a315847fa451d29ec6c3f1d7ff006fb2f87","observation_id":"b15b7d3e-923d-4d9c-a933-845dbb8d00d6","resolution":{"observed_at":"2026-08-09T11:12:09.618747Z","resolver_source":"arxiv_id_nonexistent","status":"malformed_identifier"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:15.874087Z","title":"We use varying α∈[0,1] for WiSE-FT","venue":null,"work_id":"6840fa9d-ac65-4b67-bc70-5fca232a0006","year":null},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.289178Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:996d6b598f7ec77fb486b3c4f06abc13abbba8a2538804fcfccb0a41de6ba81d","observation_id":"7329e9ca-7020-480f-a135-f2f3755f7d4e","resolution":{"observed_at":"2026-08-09T11:12:15.877828Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2109.01903","last_updated":"2022-06-21T21:50:28Z","snapshot_observed_at":"2026-07-06T11:44:19.030296Z","submitted_at":"2021-09-04T17:11:28Z","title":"Robust fine-tuning of zero-shot models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2109.01903","snapshot_observed_at":"2026-08-09T11:12:08.212220Z","title":"findings-emnlp.715/","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":715,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.212220Z"},"links":{"cited_paper":"/paper/2109.01903","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:d5eeda58a51d53e7a30de76869a8de9b4136ce80577c43e2737339ae630ee341","observation_id":"4a07ffec-cf2a-40ba-ae5f-52c075750bf2","resolution":{"observed_at":"2026-08-09T11:12:08.212220Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2108.07732","last_updated":"2021-08-16T03:57:30Z","snapshot_observed_at":"2026-08-02T19:23:53.535075Z","submitted_at":"2021-08-16T03:57:30Z","title":"Program Synthesis with Large Language Models","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2108.07732","snapshot_observed_at":"2026-08-09T11:12:08.151280Z","title":"org/CorpusID:6546520","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2015,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.151280Z"},"links":{"cited_paper":"/paper/2108.07732","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:ea31452e98690a167bde9acdcc579d30e2400d0934fb0b671d0341935df79dab","observation_id":"d2a9bcb5-33a8-432d-ac1b-f24b15ac94e8","resolution":{"observed_at":"2026-08-09T11:12:08.151280Z","resolver_source":null,"status":"malformed_identifier"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2501.05559","last_updated":"2025-01-09T20:11:08Z","snapshot_observed_at":"2026-07-06T20:18:58.432111Z","submitted_at":"2025-01-09T20:11:08Z","title":"Soup to go: mitigating forgetting during continual learning with model averaging","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2501.05559","snapshot_observed_at":"2026-08-09T11:12:08.181402Z","title":"org/CorpusID:4704285","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2016,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.181402Z"},"links":{"cited_paper":"/paper/2501.05559","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:9c7f457056cc6b396d3cbb5416d008881956adcec720f144d6078b4a3345cbe0","observation_id":"652fedc3-3e29-4a3d-ab20-63b95647d0a0","resolution":{"observed_at":"2026-08-09T11:12:08.181402Z","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":"2026-08-09T11:12:16.110267Z","title":"org/CorpusID:19243534","venue":null,"work_id":"02f18368-9b08-40bc-aed2-e275304706cf","year":2020},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2017,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.174488Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:cb15bf07e82280cc6ece3daa0019f1287280d85fb4691f8416320b11188e881b","observation_id":"669d8da3-f5fc-4c34-a3dd-a28249ef1196","resolution":{"observed_at":"2026-08-09T11:12:16.114414Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+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":"2026-08-09T11:12:16.087482Z","title":"org/CorpusID:3652214","venue":null,"work_id":"db024b0b-a598-4519-8b6f-ee8fcb5d7440","year":2024},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2018,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.216651Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:0f85fb385f67523e9042a20ec0d9bdff2b4f44c059d2bed302cdb7bcae0c138e","observation_id":"6375b629-10eb-4e90-82b5-d4b965620b82","resolution":{"observed_at":"2026-08-09T11:12:16.091009Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2406.16797","last_updated":"2024-06-25T13:46:41Z","snapshot_observed_at":"2026-08-06T08:17:30.657677Z","submitted_at":"2024-06-24T16:58:23Z","title":"Lottery Ticket Adaptation: Mitigating Destructive Interference in LLMs","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2406.16797","snapshot_observed_at":"2026-08-09T11:12:08.196583Z","title":"org/CorpusID:102353035","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2019,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.196583Z"},"links":{"cited_paper":"/paper/2406.16797","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:bcef2a400bd0d5ba17bdac3e17feb0b84d8a434d12fc2372ef6a3bf1cfdd9d70","observation_id":"bceb9997-81d9-4e03-b133-b29e78819aa3","resolution":{"observed_at":"2026-08-09T11:12:08.196583Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2402.14270","last_updated":"2024-03-01T15:21:16Z","snapshot_observed_at":"2026-08-10T04:11:56.895467Z","submitted_at":"2024-02-22T04:10:57Z","title":"Take the Bull by the Horns: Hard Sample-Reweighted Continual Training Improves LLM Generalization","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2402.14270","snapshot_observed_at":"2026-08-09T11:12:08.160998Z","title":"cc/paper_files/paper/2020/file/ aa2a77371374094fe9e0bc1de3f94ed9-Paper","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2020,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.160998Z"},"links":{"cited_paper":"/paper/2402.14270","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:5bc1349cda04577b164bc509853159e00274275dff97209b738c069967dfa2dd","observation_id":"2dc35d19-4a5f-42b8-9dae-830518ad121a","resolution":{"observed_at":"2026-08-09T11:12:08.160998Z","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":"2026-08-09T11:12:16.122255Z","title":"org/CorpusID:232427874","venue":null,"work_id":"10127318-b224-484a-80e4-2f59e156842f","year":2013},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2021,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.156140Z"},"links":{"citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:1ee40e60f3d6e5cf3c678f6f69c7a9d1718deafee6e6d25f2e29e20941468b6f","observation_id":"0975d986-e081-4f49-9190-1d3485098c4e","resolution":{"observed_at":"2026-08-09T11:12:16.126050Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1706.00043","last_updated":"2017-09-13T12:54:33Z","snapshot_observed_at":"2026-08-03T17:24:12.244410Z","submitted_at":"2017-05-31T18:25:09Z","title":"Biased Importance Sampling for Deep Neural Network Training","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1706.00043","snapshot_observed_at":"2026-08-09T11:12:08.177819Z","title":"Katharopoulos, A","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2022,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.177819Z"},"links":{"cited_paper":"/paper/1706.00043","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:fa068e2d70ab61555f570c212cf406e05705d1f21a9d5a0eb5f15874e8cd8ea5","observation_id":"26d3063b-2bd8-4fac-8c5d-c7760eb26654","resolution":{"observed_at":"2026-08-09T11:12:08.177819Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.06825","last_updated":"2023-10-10T17:54:58Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2023-10-10T17:54:58Z","title":"Mistral 7B","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.06825","snapshot_observed_at":"2026-08-09T11:12:08.170193Z","title":"URL https: //doi.org/10.48550/arXiv.2310.06825","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting","version":2},"reference_index":2023,"source":"pdf_text","source_observed_at":"2026-08-09T11:12:08.170193Z"},"links":{"cited_paper":"/paper/2310.06825","citing_paper":"/paper/2502.02797"},"observation_digest":"sha256:291d96b703ba691333bfdd46bec1898524aedbcb12a232c156ded8a1eb8c913d","observation_id":"b50c536a-cea3-42cb-bcb6-2a261be1382f","resolution":{"observed_at":"2026-08-09T11:12:08.170193Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2502.02797","last_updated":"2025-06-12T03:04:07Z","latest_version":2,"primary_category":"cs.LG","snapshot_observed_at":"2026-08-10T12:12:06.267872Z","submitted_at":"2025-02-05T00:49:59Z","title":"Upweighting Easy Samples in Fine-Tuning Mitigates Forgetting"},"reference_resolution":{"displayed":38,"state_counts":{"malformed_identifier":3,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":18,"verified_exact":2,"verified_fuzzy":15},"total_outbound_references":38},"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":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"thesis":"As of 10 August 2026, this Paper Citation Record lists 38 of 38 outbound references and 4 inbound Pith citation observations for arXiv:2502.02797."}