{"as_of":"2026-08-21T23:04:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:235bdf36bcb52b56d2a668668a8152bc03c74b85bf9f0cc55e2a5f137c309b92","coverage":[{"denominator":34,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":34,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-11T17:11:02.360439Z","state":"measured"},{"denominator":35,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":35,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-21T06:32:19.484+00:00","state":"measured"},{"denominator":1,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":1,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-06T13:37:54.934359Z","state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":1,"source":"pith","source_observed_at":"2026-08-10T05:30:23.456663Z","state":"measured"}],"external_citation_measurements":[{"count":0,"observed_at":"2026-08-10T05:30:23.456663Z","source":"pith"}],"inbound":[{"citation":{"cited_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"cited_work":{"arxiv_id":"2412.09263","doi":"10.48550/arxiv.2412.09263","metadata_source":"pith","pith_arxiv_id":"2412.09263","snapshot_observed_at":"2026-08-10T05:30:23.456663Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","venue":"cs.CL","work_id":"8101b417-2b82-44eb-a441-b4feef34ab86","year":2024},"citing_paper":{"arxiv_id":"2507.20419","last_updated":"2025-07-27T21:30:50Z","snapshot_observed_at":"2026-08-14T23:24:33.154791Z","submitted_at":"2025-07-27T21:30:50Z","title":"Survey of NLU Benchmarks Diagnosing Linguistic Phenomena: Why not Standardize Diagnostics Benchmarks?","version":1},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-06T13:37:54.934359Z"},"links":{"cited_paper":"/paper/2412.09263","citing_paper":"/paper/2507.20419"},"observation_digest":"sha256:4ec407947e97d27414ed4468818bf77874e1360daecf3ba9a694a148b4aa10e5","observation_id":"3ff88a52-1719-405a-80ac-5075153207c8","resolution":{"observed_at":"2026-08-06T13:38:04.472952Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+00:00","source":"retraction_watch"}],"state":"measured"}}],"links":{"evidence":"/evidence","html":"/paper/2412.09263/citation-record","integrity":"/paper/2412.09263/integrity","json":"/paper/2412.09263/citation-record.json","paper":"/paper/2412.09263"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-11T17:11:17.980388Z","title":"A large annotated corpus for learning natural language inference,","venue":null,"work_id":"a14a397c-59ed-4ffe-82fb-97bd89ad027b","year":2015},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.193107Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:7162c47aac4ca684de56577dc7f5ffa69c00b72c3c309dc07283af0680f5157d","observation_id":"c3362e41-ea9d-4660-8b9d-47661238666c","resolution":{"observed_at":"2026-08-11T17:11:17.985756Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.964554Z","title":"A decomposable attention model for natural language inference,","venue":null,"work_id":"fa89df9e-fe8c-41aa-a8ab-d5ef3e1215d9","year":2016},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.199774Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:14e0209a111a9dbb1349f3be398fc58f37172be5efc13bebec6c2c661e8a4e26","observation_id":"e859b954-e64f-42ab-969b-131b8afd1b0e","resolution":{"observed_at":"2026-08-11T17:11:17.969552Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.948595Z","title":"Enhanced lstm for natural language inference,","venue":null,"work_id":"d01d6fae-2007-4787-865f-d24c7a06c113","year":2017},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.204883Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:438d8c2d1dbedfa4b8b0e9bdca98d184a49161484e9add1f042711ac558e2637","observation_id":"f9e522f1-6741-4a7c-9d7c-077ee6e1c0c7","resolution":{"observed_at":"2026-08-11T17:11:17.953895Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.932661Z","title":"Improving language understanding by generative pre-training,","venue":null,"work_id":"389d937f-204a-4c80-a14f-a92c34c7fd28","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.209956Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:53acab7e96bfb3979c17ea749888f7cbf86a86af0d5d329846570af3831daad7","observation_id":"de836a0f-e1e6-4321-8efd-c24290ec72cc","resolution":{"observed_at":"2026-08-11T17:11:17.937895Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:02.216976Z","title":"Bert: Pre-training of deep bidirectional transformers for language understanding,","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.216976Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:496b0dfb0233b941d0b085eb2bedcd95daef801ec44c9fecaf93b752623666ca","observation_id":"f6c3f9eb-9837-4c7c-8298-3d0c54b46588","resolution":{"observed_at":"2026-08-11T17:11:02.216976Z","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":"2026-08-11T17:11:17.906696Z","title":"Xlnet: General- ized autoregressive pretraining for language understanding,","venue":null,"work_id":"adb81582-8601-408f-a51c-1eaaea649496","year":2019},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.222518Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:c881546169e51b22df9283ce1a1480ee50bccdeccde0812aa791658612fdcf93","observation_id":"d1bcf9c1-ab9b-411a-97bd-dd686767ce79","resolution":{"observed_at":"2026-08-11T17:11:17.911632Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1907.11692","last_updated":"2019-07-26T17:48:29Z","snapshot_observed_at":"2026-08-16T14:33:50.657682Z","submitted_at":"2019-07-26T17:48:29Z","title":"RoBERTa: A Robustly Optimized BERT Pretraining Approach","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1907.11692","snapshot_observed_at":"2026-08-11T17:11:02.228497Z","title":"Roberta: A robustly optimized bert pretraining approach,","venue":null,"work_id":null,"year":1907},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.228497Z"},"links":{"cited_paper":"/paper/1907.11692","citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:f500e41633475de22476f072c8301d157f962c1eb623d86095c7e04097aab583","observation_id":"3f3828b2-5f33-46f0-bbb2-d0ab5ad84316","resolution":{"observed_at":"2026-08-11T17:11:02.228497Z","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":"2026-08-11T17:11:17.891252Z","title":"Albert: A lite bert for self-supervised learning of language representations,","venue":null,"work_id":"1578e236-0598-455e-ae52-241082ba8765","year":2020},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.233672Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:4d7732003f8e20548caa779bd5dd8de493102944681213365843905b255636c4","observation_id":"b574fda0-445c-4e26-82ed-62311d1e348e","resolution":{"observed_at":"2026-08-11T17:11:17.896105Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2106.03654","last_updated":"2021-06-07T14:33:25Z","snapshot_observed_at":"2026-08-21T05:22:16.187304Z","submitted_at":"2021-06-07T14:33:25Z","title":"The Convexity and Concavity of Envelopes of the Minimum-Relative-Entropy Region for the DSBS","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2106.03654","snapshot_observed_at":"2026-08-11T17:11:02.238436Z","title":"Deberta: Decoding-enhanced bert with disentangled attention,","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.238436Z"},"links":{"cited_paper":"/paper/2106.03654","citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:433b300b3a5ec4bd881ad2841b121baa9fffe0292cd5ce0d0038f08c41db4149","observation_id":"ab741dd5-9ff5-4bc8-801a-a97f02c7cd13","resolution":{"observed_at":"2026-08-11T17:11:02.238436Z","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":"2026-08-11T17:11:17.875333Z","title":"Entailment as few-shot learner,","venue":null,"work_id":"ec79f787-c9f6-451f-bb6f-da86f5c0b78b","year":2021},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.243349Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:7207389f9c936b91dcd6b3cf94c3edabb1e3f1820fba5159a1c670c472920ca5","observation_id":"749b8d35-7ef8-4903-9fb7-921540c3cbb3","resolution":{"observed_at":"2026-08-11T17:11:17.880142Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.859533Z","title":"Papers with Code - SNLI Benchmark (Natural Language Inference) — paperswithcode.com","venue":null,"work_id":"84ed8ee1-0eef-4e82-a9ba-9c92ca8046c8","year":2024},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.248603Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:ec8bc1851103268c04066b3b93901f6d46b466249289453c933c9ae0a00cb192","observation_id":"db737871-18a3-4f5a-aa2e-e59560182cfd","resolution":{"observed_at":"2026-08-11T17:11:17.864886Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.842871Z","title":"An extended model of natural logic,","venue":null,"work_id":"c1f997dd-1e21-4fee-b985-a7b617974be1","year":2009},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.253577Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:9a173eb94da6499a70190c927b088b59226174b28156beff4661a9246ef5fb3a","observation_id":"4ae4b648-606b-437b-aed1-897cd2b083ee","resolution":{"observed_at":"2026-08-11T17:11:17.848447Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.826764Z","title":"Natural language inference,","venue":null,"work_id":"f151766d-f403-4d3c-969d-44f613856d26","year":2008},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.258774Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:78e09a68799792a8a376f7cdfe4fec928f23ecd3b30b53703c79e221c468a7ca","observation_id":"479ac25a-b80a-4e5b-88d0-b1c4a2881874","resolution":{"observed_at":"2026-08-11T17:11:17.831902Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.810849Z","title":"Deep contextualized word representations,","venue":null,"work_id":"a9ecd9b6-1c63-45a1-a3bb-e3b96e287fa7","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.263664Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:e6673a28ed71f883fc185b317d2ceb1ba766d06730f55787b051099f9b137f4f","observation_id":"83d902cd-999b-415f-ade6-94c131c6bd2a","resolution":{"observed_at":"2026-08-11T17:11:17.815699Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.795836Z","title":"Llms will always hallucinate, and we need to live with this,","venue":null,"work_id":"52b6bc26-7d3b-44eb-a618-1031d48bfeb8","year":2024},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.268885Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:4057b614d9beb61a2c085fb71dcd2a474225a9558d022a677778ca30c60f2234","observation_id":"e0d73e51-661e-491f-ad63-ead84766aca8","resolution":{"observed_at":"2026-08-11T17:11:17.800711Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.779168Z","title":"A broad-coverage challenge corpus for sentence understanding through inference,","venue":null,"work_id":"6bf06995-d5c9-4fe8-8005-5e6a6a9b85e8","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.273652Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:e1ccb5abd329f26caa8bab29d748080b91eb7469ebbfbfd71cab98103ba0ef03","observation_id":"e8db2c24-3380-4c76-82c4-7c324335f226","resolution":{"observed_at":"2026-08-11T17:11:17.784528Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.762105Z","title":"Exploring the limits of transfer learning with a unified text-to-text transformer,","venue":null,"work_id":"890a05cc-ff29-4291-b734-e7f71d92fcaa","year":2023},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.278349Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:81dbc3cac2076d5ad44f0251283e08556287be03f9b976fbacfd3c6caa728a21","observation_id":"99245431-8aeb-459b-a8fe-074b8c0e9c2a","resolution":{"observed_at":"2026-08-11T17:11:17.767277Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.746348Z","title":"Large dual encoders are generalizable retrievers,","venue":null,"work_id":"db9b4234-8f42-4c0f-a7a6-3fec3118060a","year":2021},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.283359Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:db61b98c989c0a35d796b904bcb78f95edf73430e866a83766ede73036c90c16","observation_id":"efe4a68a-ff10-4b33-9f5d-2759f7c15099","resolution":{"observed_at":"2026-08-11T17:11:17.751376Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.730172Z","title":"The pascal recognising textual entailment challenge,","venue":null,"work_id":"c9e6afa5-ce98-4c95-b772-62ce80243921","year":2005},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.288194Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:8fc91028f3889dec2eccc4f66961d58a83d87194b50be6048768ccbbf112bf09","observation_id":"1f237836-d47c-4908-bf85-0a0e696ac808","resolution":{"observed_at":"2026-08-11T17:11:17.735785Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.714499Z","title":"Knowledgeable reader: Enhancing cloze-style reading comprehension with external commonsense knowledge,","venue":null,"work_id":"567f952c-8a20-49a6-b31b-39a5c39ea747","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.293075Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:677b5f7c42272f986ed81b81e23c60b7a414e359933f18e8a259311b2ead370e","observation_id":"32fbac64-e47c-422b-87a7-30a4b3943e57","resolution":{"observed_at":"2026-08-11T17:11:17.719550Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.699216Z","title":"Xnli: Evaluating cross-lingual sentence representations,","venue":null,"work_id":"eece6022-0064-4311-a83d-c035cf533c99","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.298048Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:2959569109dfd7b5b7ecb99c3db47fa2f7286a9bb08ab8339cfddf2400c6f0f9","observation_id":"ee12c8cc-3111-4b88-b496-85f229e711eb","resolution":{"observed_at":"2026-08-11T17:11:17.703903Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:02.302897Z","title":"Attention is all you need,","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.302897Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:70ef326ae3bb88acf47466e042a6f4d3121b1c25ce52b32023fcb3805492b259","observation_id":"50f8ad58-6d3b-4a1b-8664-7d0b44558d3d","resolution":{"observed_at":"2026-08-11T17:11:02.302897Z","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":"2026-08-11T17:11:17.673119Z","title":"High-precision medical speech recognition through synthetic data and semantic correction: United-medasr,","venue":null,"work_id":"93d072ee-095a-4063-8779-18003c09325a","year":2024},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.307613Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:10a6d43852ee51caaa1c0d1aa06e50fe6b2b2aca243631bdfd446d03dace475a","observation_id":"fcfd8e92-097a-433b-8ae6-df5a3cac39b7","resolution":{"observed_at":"2026-08-11T17:11:17.678397Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.657757Z","title":"Bart: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension,","venue":null,"work_id":"4b9004ff-97b0-489e-bfad-895bf2001e8c","year":2019},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.312585Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:a89cefd18a5c20792b6fa23a2a0f00dcd082b904fe02d79ef91f691fe7ca9d58","observation_id":"ddf00858-acaa-4171-9928-ecc0a3b5c45b","resolution":{"observed_at":"2026-08-11T17:11:17.662919Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.641124Z","title":"Smart: Robust and efficient fine- tuning for pre-trained natural language models through principled regularized optimization,","venue":null,"work_id":"4197db3c-d8e6-4de8-af09-b19b6b1cec64","year":2020},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.317246Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:53d609f591ade61d77c7af427cf4ec08e94970682780cc38de59d39be7811dff","observation_id":"945e4c18-fcf6-4061-a079-4c9a90f4660c","resolution":{"observed_at":"2026-08-11T17:11:17.646062Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.625774Z","title":"Palm 2 technical report,","venue":null,"work_id":"c7cffbfb-da4f-4b86-afc8-decd4299f604","year":2023},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.322324Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:b4bd15111cb75131bf3e520d555d0bbd539761cb2ef61eeaf230d07ae6eaef5f","observation_id":"f9e001c5-b50c-4248-a181-5422d1f0d53a","resolution":{"observed_at":"2026-08-11T17:11:17.630567Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.610057Z","title":"Claude 3 model card,","venue":null,"work_id":"68a8c87c-b9e4-4366-8367-3e222264e2a7","year":2023},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.327713Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:1298e3a562bbae6bbfdada8ce92ba03b7ef58a122aca450998cfb4006c46bd89","observation_id":"28e69e27-d56f-40b2-98a7-7ba607667fd4","resolution":{"observed_at":"2026-08-11T17:11:17.615450Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.593099Z","title":"Gpt-4 technical report,","venue":null,"work_id":"9399aa81-2e73-4209-8caf-21a18975de88","year":2024},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.332739Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:3083cd9d42e7cfe2a686669567fbfdea8dc28e77f7da59e5bc8e1676a2046fdb","observation_id":"11ce0e6e-c612-41e7-8367-d6febd490e1d","resolution":{"observed_at":"2026-08-11T17:11:17.599349Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.576302Z","title":"Scitail: A textual entailment dataset from science question answering,","venue":null,"work_id":"1fc1127a-8154-4c42-8c41-e7daf4f0f097","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.338565Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:c467ddcc0ac77faa335af83ee5f77fe90ecf8d0b57ae838c7e911f4cdee4767c","observation_id":"37785f1c-5ba9-41cb-9f94-2a0ec59ae92b","resolution":{"observed_at":"2026-08-11T17:11:17.581490Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.560100Z","title":"Multi-task deep neural networks for natural language un- derstanding,","venue":null,"work_id":"85316c43-00a0-4373-a8d8-13489e315f0d","year":2019},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.342738Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:b7ef08bef92f293a5cda510745085657dbfde2b4811276ccecf3b4c65a0f42ee","observation_id":"67a28d43-cd13-4103-98c5-fe86b68b2973","resolution":{"observed_at":"2026-08-11T17:11:17.564962Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":"2004.07828","doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-11T17:11:17.454163Z","title":"Anli: A new benchmark for natural language understanding,","venue":null,"work_id":"ac8cf586-0c83-4020-827a-ce7657131bc8","year":2004},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.347460Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:a30d4d36e089ae2bae9d8fd7eb7d034a09cad218b3bb5e402487d793e9f8ab94","observation_id":"df94679d-53c7-410e-8915-4578e1fe08ea","resolution":{"observed_at":"2026-08-11T17:11:17.464500Z","resolver_source":"raw_fallback","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.544646Z","title":"e-snli: Natural language inference with natural language explanations,","venue":null,"work_id":"b5e911c4-9125-4d0f-a8ff-52633ce607de","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.351835Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:748051efa118ab4ca6c982a18598f4016ed260a360b6557dff15ed12bfaf8d2f","observation_id":"c1b1ed91-b2f6-4dc0-9c9e-ceb90a826333","resolution":{"observed_at":"2026-08-11T17:11:17.549331Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.528725Z","title":"Scaling instruction-finetuned language models,","venue":null,"work_id":"1868c356-c226-441c-a568-3dd89bf914c6","year":2022},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.356075Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:8179bcba8422aa44d83aa91dfda04d10796cc99b3a25c96e69fef3f1310dc05e","observation_id":"ff5fe36f-63df-4a93-8873-9721a1c47270","resolution":{"observed_at":"2026-08-11T17:11:17.534048Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+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":"2026-08-11T17:11:17.511399Z","title":"e-snli: Natural language inference with natural language explanations,","venue":null,"work_id":"c09c2cc9-aa4c-426c-9b2a-e0e7e4b45ede","year":2018},"citing_paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI","version":2},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-08-11T17:11:02.360439Z"},"links":{"citing_paper":"/paper/2412.09263"},"observation_digest":"sha256:c2298dd850565033d957a2db4801eb8d4a88b24d6ae05362b5bb8ba82ee60d57","observation_id":"43419b01-a914-46b3-8fc2-04fbb1196363","resolution":{"observed_at":"2026-08-11T17:11:17.517682Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+00:00","source":"retraction_watch"}],"state":"measured"}}],"paper":{"arxiv_id":"2412.09263","last_updated":"2024-12-13T06:28:11Z","latest_version":2,"primary_category":"cs.CL","snapshot_observed_at":"2026-08-11T17:03:32.493472Z","submitted_at":"2024-12-12T13:21:09Z","title":"First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI"},"reference_resolution":{"displayed":34,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":4,"verified_exact":1,"verified_fuzzy":29},"total_outbound_references":34},"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":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+00:00","source":"retraction_watch"}],"thesis":"As of 21 August 2026, this Paper Citation Record lists 34 of 34 outbound references and 1 inbound Pith citation observation for arXiv:2412.09263."}