{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XOCHYLDH6CVNLFY6YXDB7IBSHR","short_pith_number":"pith:XOCHYLDH","schema_version":"1.0","canonical_sha256":"bb847c2c67f0aad5971ec5c61fa0323c5219117cf75e13c9ea1fe1ec52d12f0e","source":{"kind":"arxiv","id":"2602.15136","version":2},"attestation_state":"computed","paper":{"title":"Universal priors: solving empirical Bayes via Bayesian inference and pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Anzo Teh, Nick Cannella, Yanjun Han, Yury Polyanskiy","submitted_at":"2026-02-16T19:29:27Z","abstract_excerpt":"We theoretically justify the recent empirical finding of [Teh et al., 2025] that a transformer pretrained on synthetically generated data achieves strong performance on empirical Bayes (EB) problems. We take an indirect approach to this question: rather than analyzing the model architecture or training dynamics, we ask why a pretrained Bayes estimator, trained under a prespecified training distribution, can adapt to arbitrary test distributions. Focusing on Poisson EB problems, we identify the existence of universal priors such that training under these priors yields a near-optimal regret boun"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2602.15136","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-02-16T19:29:27Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"56660222c8dc9482423846d60fba6619cfaf6b19973def41e6974cc70cec2947","abstract_canon_sha256":"2831a08628e878bab0a1ff22940da2beeaa63526899ddd1e650d3703ab868f95"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:13:54.488439Z","signature_b64":"vSEEKXRkhbBflsC/QKh7I12/0hiCLz0IQblM7nA028cIgI7tghoKw64AiqIHT2ZwF4CTBap25bdlD0h38CJMCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bb847c2c67f0aad5971ec5c61fa0323c5219117cf75e13c9ea1fe1ec52d12f0e","last_reissued_at":"2026-06-23T03:13:54.487947Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:13:54.487947Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Universal priors: solving empirical Bayes via Bayesian inference and pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Anzo Teh, Nick Cannella, Yanjun Han, Yury Polyanskiy","submitted_at":"2026-02-16T19:29:27Z","abstract_excerpt":"We theoretically justify the recent empirical finding of [Teh et al., 2025] that a transformer pretrained on synthetically generated data achieves strong performance on empirical Bayes (EB) problems. We take an indirect approach to this question: rather than analyzing the model architecture or training dynamics, we ask why a pretrained Bayes estimator, trained under a prespecified training distribution, can adapt to arbitrary test distributions. Focusing on Poisson EB problems, we identify the existence of universal priors such that training under these priors yields a near-optimal regret boun"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.15136","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2602.15136/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2602.15136","created_at":"2026-06-23T03:13:54.488005+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.15136v2","created_at":"2026-06-23T03:13:54.488005+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.15136","created_at":"2026-06-23T03:13:54.488005+00:00"},{"alias_kind":"pith_short_12","alias_value":"XOCHYLDH6CVN","created_at":"2026-06-23T03:13:54.488005+00:00"},{"alias_kind":"pith_short_16","alias_value":"XOCHYLDH6CVNLFY6","created_at":"2026-06-23T03:13:54.488005+00:00"},{"alias_kind":"pith_short_8","alias_value":"XOCHYLDH","created_at":"2026-06-23T03:13:54.488005+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2606.21707","citing_title":"Quasi-Bayes empirical Bayes estimation of sums of random variables","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2607.02340","citing_title":"Merging of Bayes and quasi-Bayes empirical Bayes procedures for Poisson compound decisions","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2605.03594","citing_title":"Poisson Empirical Bayes via Gamma-Smoothed Nonparametric Maximum Likelihood","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR","json":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR.json","graph_json":"https://pith.science/api/pith-number/XOCHYLDH6CVNLFY6YXDB7IBSHR/graph.json","events_json":"https://pith.science/api/pith-number/XOCHYLDH6CVNLFY6YXDB7IBSHR/events.json","paper":"https://pith.science/paper/XOCHYLDH"},"agent_actions":{"view_html":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR","download_json":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR.json","view_paper":"https://pith.science/paper/XOCHYLDH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.15136&json=true","fetch_graph":"https://pith.science/api/pith-number/XOCHYLDH6CVNLFY6YXDB7IBSHR/graph.json","fetch_events":"https://pith.science/api/pith-number/XOCHYLDH6CVNLFY6YXDB7IBSHR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR/action/storage_attestation","attest_author":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR/action/author_attestation","sign_citation":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR/action/citation_signature","submit_replication":"https://pith.science/pith/XOCHYLDH6CVNLFY6YXDB7IBSHR/action/replication_record"}},"created_at":"2026-06-23T03:13:54.488005+00:00","updated_at":"2026-06-23T03:13:54.488005+00:00"}