{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:VJDODCYW7PSSDYDGKQ4JUNHAIT","short_pith_number":"pith:VJDODCYW","schema_version":"1.0","canonical_sha256":"aa46e18b16fbe521e06654389a34e044d9ccba8288e067ccbce403948ef4fd95","source":{"kind":"arxiv","id":"2109.09193","version":1},"attestation_state":"computed","paper":{"title":"Towards Zero-Label Language Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Adams Wei Yu, Orhan Firat, Yuan Cao, Zirui Wang","submitted_at":"2021-09-19T19:00:07Z","abstract_excerpt":"This paper explores zero-label learning in Natural Language Processing (NLP), whereby no human-annotated data is used anywhere during training and models are trained purely on synthetic data. At the core of our framework is a novel approach for better leveraging the powerful pretrained language models. Specifically, inspired by the recent success of few-shot inference on GPT-3, we present a training data creation procedure named Unsupervised Data Generation (UDG), which leverages few-shot prompts to synthesize high-quality training data without real human annotations. Our method enables zero-l"},"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":"2109.09193","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-09-19T19:00:07Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"832940bd2105de37b27124dfff6019eb98ff63bb7015521ea86eb6baa72e7caa","abstract_canon_sha256":"22334a020d36951f1004b041cbf8d20d3f37a17060f348bbe87d513d3ec141eb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:15:42.596637Z","signature_b64":"sCrBUf2INH1D8SktudN+xYagwGIVmLhO++mNV+LWprrW9Zi5NixZs6yt7HLZEP7d5jWWtix3ZKwJ+NiL1rnhDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa46e18b16fbe521e06654389a34e044d9ccba8288e067ccbce403948ef4fd95","last_reissued_at":"2026-07-05T03:15:42.596133Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:15:42.596133Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Zero-Label Language Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Adams Wei Yu, Orhan Firat, Yuan Cao, Zirui Wang","submitted_at":"2021-09-19T19:00:07Z","abstract_excerpt":"This paper explores zero-label learning in Natural Language Processing (NLP), whereby no human-annotated data is used anywhere during training and models are trained purely on synthetic data. At the core of our framework is a novel approach for better leveraging the powerful pretrained language models. Specifically, inspired by the recent success of few-shot inference on GPT-3, we present a training data creation procedure named Unsupervised Data Generation (UDG), which leverages few-shot prompts to synthesize high-quality training data without real human annotations. Our method enables zero-l"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.09193","kind":"arxiv","version":1},"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/2109.09193/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":"2109.09193","created_at":"2026-07-05T03:15:42.596184+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.09193v1","created_at":"2026-07-05T03:15:42.596184+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.09193","created_at":"2026-07-05T03:15:42.596184+00:00"},{"alias_kind":"pith_short_12","alias_value":"VJDODCYW7PSS","created_at":"2026-07-05T03:15:42.596184+00:00"},{"alias_kind":"pith_short_16","alias_value":"VJDODCYW7PSSDYDG","created_at":"2026-07-05T03:15:42.596184+00:00"},{"alias_kind":"pith_short_8","alias_value":"VJDODCYW","created_at":"2026-07-05T03:15:42.596184+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2201.11990","citing_title":"Using DeepSpeed and Megatron to Train Megatron-Turing NLG 530B, A Large-Scale Generative Language Model","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2409.12917","citing_title":"Training Language Models to Self-Correct via Reinforcement Learning","ref_index":113,"is_internal_anchor":false},{"citing_arxiv_id":"2309.00267","citing_title":"RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback","ref_index":106,"is_internal_anchor":false},{"citing_arxiv_id":"2112.04359","citing_title":"Ethical and social risks of harm from Language Models","ref_index":288,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18939","citing_title":"TabEmb: Joint Semantic-Structure Embedding for Table Annotation","ref_index":16,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT","json":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT.json","graph_json":"https://pith.science/api/pith-number/VJDODCYW7PSSDYDGKQ4JUNHAIT/graph.json","events_json":"https://pith.science/api/pith-number/VJDODCYW7PSSDYDGKQ4JUNHAIT/events.json","paper":"https://pith.science/paper/VJDODCYW"},"agent_actions":{"view_html":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT","download_json":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT.json","view_paper":"https://pith.science/paper/VJDODCYW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.09193&json=true","fetch_graph":"https://pith.science/api/pith-number/VJDODCYW7PSSDYDGKQ4JUNHAIT/graph.json","fetch_events":"https://pith.science/api/pith-number/VJDODCYW7PSSDYDGKQ4JUNHAIT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT/action/storage_attestation","attest_author":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT/action/author_attestation","sign_citation":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT/action/citation_signature","submit_replication":"https://pith.science/pith/VJDODCYW7PSSDYDGKQ4JUNHAIT/action/replication_record"}},"created_at":"2026-07-05T03:15:42.596184+00:00","updated_at":"2026-07-05T03:15:42.596184+00:00"}