{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:LDBKU3KIVCBEEDNHKWAZ3RLPXQ","short_pith_number":"pith:LDBKU3KI","schema_version":"1.0","canonical_sha256":"58c2aa6d48a882420da755819dc56fbc357691ba329048c56125df5a577f18ee","source":{"kind":"arxiv","id":"2110.07752","version":2},"attestation_state":"computed","paper":{"title":"Hindsight: Posterior-guided training of retrievers for improved open-ended generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Ashwin Paranjape, Christopher D. Manning, Christopher Potts, Matei Zaharia, Omar Khattab","submitted_at":"2021-10-14T22:24:57Z","abstract_excerpt":"Many text generation systems benefit from using a retriever to retrieve passages from a textual knowledge corpus (e.g., Wikipedia) which are then provided as additional context to the generator. For open-ended generation tasks (like generating informative utterances in conversations) many varied passages may be equally relevant and we find that existing methods that jointly train the retriever and generator underperform: the retriever may not find relevant passages even amongst the top-10 and hence the generator may not learn a preference to ground its generated output in them. We propose usin"},"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":"2110.07752","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-10-14T22:24:57Z","cross_cats_sorted":["cs.IR"],"title_canon_sha256":"56e0824a09f9e1320e4e72286ff094af65a989666f8c6d94bbdbb80997bf5c27","abstract_canon_sha256":"fc424671dc0cc2237236355974719b5654f599b45b9c3748a62ce10af25081e9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:24:31.916116Z","signature_b64":"H7FFvlOOqgPUvNcqBqTVWlCxAQgqahWF2r9CHnDiaVqBtHSElPsLYzONd5BlPoYswiUQvi8McRwDF6CQooJxAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"58c2aa6d48a882420da755819dc56fbc357691ba329048c56125df5a577f18ee","last_reissued_at":"2026-07-05T03:24:31.915641Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:24:31.915641Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hindsight: Posterior-guided training of retrievers for improved open-ended generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Ashwin Paranjape, Christopher D. Manning, Christopher Potts, Matei Zaharia, Omar Khattab","submitted_at":"2021-10-14T22:24:57Z","abstract_excerpt":"Many text generation systems benefit from using a retriever to retrieve passages from a textual knowledge corpus (e.g., Wikipedia) which are then provided as additional context to the generator. For open-ended generation tasks (like generating informative utterances in conversations) many varied passages may be equally relevant and we find that existing methods that jointly train the retriever and generator underperform: the retriever may not find relevant passages even amongst the top-10 and hence the generator may not learn a preference to ground its generated output in them. We propose usin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2110.07752","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/2110.07752/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":"2110.07752","created_at":"2026-07-05T03:24:31.915703+00:00"},{"alias_kind":"arxiv_version","alias_value":"2110.07752v2","created_at":"2026-07-05T03:24:31.915703+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2110.07752","created_at":"2026-07-05T03:24:31.915703+00:00"},{"alias_kind":"pith_short_12","alias_value":"LDBKU3KIVCBE","created_at":"2026-07-05T03:24:31.915703+00:00"},{"alias_kind":"pith_short_16","alias_value":"LDBKU3KIVCBEEDNH","created_at":"2026-07-05T03:24:31.915703+00:00"},{"alias_kind":"pith_short_8","alias_value":"LDBKU3KI","created_at":"2026-07-05T03:24:31.915703+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2208.03299","citing_title":"Atlas: Few-shot Learning with Retrieval Augmented Language Models","ref_index":230,"is_internal_anchor":false},{"citing_arxiv_id":"2208.03299","citing_title":"Atlas: Few-shot Learning with Retrieval Augmented Language Models","ref_index":32,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ","json":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ.json","graph_json":"https://pith.science/api/pith-number/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/graph.json","events_json":"https://pith.science/api/pith-number/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/events.json","paper":"https://pith.science/paper/LDBKU3KI"},"agent_actions":{"view_html":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ","download_json":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ.json","view_paper":"https://pith.science/paper/LDBKU3KI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2110.07752&json=true","fetch_graph":"https://pith.science/api/pith-number/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/graph.json","fetch_events":"https://pith.science/api/pith-number/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/action/storage_attestation","attest_author":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/action/author_attestation","sign_citation":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/action/citation_signature","submit_replication":"https://pith.science/pith/LDBKU3KIVCBEEDNHKWAZ3RLPXQ/action/replication_record"}},"created_at":"2026-07-05T03:24:31.915703+00:00","updated_at":"2026-07-05T03:24:31.915703+00:00"}