{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:FE2LPMS5NCSOI56IZXXE7Y6AEZ","short_pith_number":"pith:FE2LPMS5","schema_version":"1.0","canonical_sha256":"2934b7b25d68a4e477c8cdee4fe3c0267ccb3d31035455b5241e90598a0cd388","source":{"kind":"arxiv","id":"2101.00438","version":2},"attestation_state":"computed","paper":{"title":"Few-Shot Question Answering by Pretraining Span Selection","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amir Globerson, Jonathan Berant, Omer Levy, Ori Ram, Yuval Kirstain","submitted_at":"2021-01-02T11:58:44Z","abstract_excerpt":"In several question answering benchmarks, pretrained models have reached human parity through fine-tuning on an order of 100,000 annotated questions and answers. We explore the more realistic few-shot setting, where only a few hundred training examples are available, and observe that standard models perform poorly, highlighting the discrepancy between current pretraining objectives and question answering. We propose a new pretraining scheme tailored for question answering: recurring span selection. Given a passage with multiple sets of recurring spans, we mask in each set all recurring spans b"},"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":"2101.00438","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2021-01-02T11:58:44Z","cross_cats_sorted":[],"title_canon_sha256":"d4be996735421640bee65edebb9ca09fada756aea493b24181475bea0f640880","abstract_canon_sha256":"26ce45cb6caab456d4bde244ef70ad0de2a0801bc40cec77d1de99c74fbc3b23"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:45:35.725419Z","signature_b64":"k9466zQcyFw5lpHigPqtF8gh641dylfQ4x6/9GS0/iORK2WYJhqnTdvkO6KjBJjumNXiDx79KAUPwrMUSpgBBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2934b7b25d68a4e477c8cdee4fe3c0267ccb3d31035455b5241e90598a0cd388","last_reissued_at":"2026-07-05T02:45:35.724925Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:45:35.724925Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Few-Shot Question Answering by Pretraining Span Selection","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amir Globerson, Jonathan Berant, Omer Levy, Ori Ram, Yuval Kirstain","submitted_at":"2021-01-02T11:58:44Z","abstract_excerpt":"In several question answering benchmarks, pretrained models have reached human parity through fine-tuning on an order of 100,000 annotated questions and answers. We explore the more realistic few-shot setting, where only a few hundred training examples are available, and observe that standard models perform poorly, highlighting the discrepancy between current pretraining objectives and question answering. We propose a new pretraining scheme tailored for question answering: recurring span selection. Given a passage with multiple sets of recurring spans, we mask in each set all recurring spans b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2101.00438","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/2101.00438/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":"2101.00438","created_at":"2026-07-05T02:45:35.724984+00:00"},{"alias_kind":"arxiv_version","alias_value":"2101.00438v2","created_at":"2026-07-05T02:45:35.724984+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2101.00438","created_at":"2026-07-05T02:45:35.724984+00:00"},{"alias_kind":"pith_short_12","alias_value":"FE2LPMS5NCSO","created_at":"2026-07-05T02:45:35.724984+00:00"},{"alias_kind":"pith_short_16","alias_value":"FE2LPMS5NCSOI56I","created_at":"2026-07-05T02:45:35.724984+00:00"},{"alias_kind":"pith_short_8","alias_value":"FE2LPMS5","created_at":"2026-07-05T02:45:35.724984+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.19482","citing_title":"Pre-training, Fine-tuning and Re-ranking: A Three-Stage Framework for Legal Question Answering","ref_index":9,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ","json":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ.json","graph_json":"https://pith.science/api/pith-number/FE2LPMS5NCSOI56IZXXE7Y6AEZ/graph.json","events_json":"https://pith.science/api/pith-number/FE2LPMS5NCSOI56IZXXE7Y6AEZ/events.json","paper":"https://pith.science/paper/FE2LPMS5"},"agent_actions":{"view_html":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ","download_json":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ.json","view_paper":"https://pith.science/paper/FE2LPMS5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2101.00438&json=true","fetch_graph":"https://pith.science/api/pith-number/FE2LPMS5NCSOI56IZXXE7Y6AEZ/graph.json","fetch_events":"https://pith.science/api/pith-number/FE2LPMS5NCSOI56IZXXE7Y6AEZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ/action/storage_attestation","attest_author":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ/action/author_attestation","sign_citation":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ/action/citation_signature","submit_replication":"https://pith.science/pith/FE2LPMS5NCSOI56IZXXE7Y6AEZ/action/replication_record"}},"created_at":"2026-07-05T02:45:35.724984+00:00","updated_at":"2026-07-05T02:45:35.724984+00:00"}