{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:4JKA73VINHKZHZ4VE5QT56LPL6","short_pith_number":"pith:4JKA73VI","schema_version":"1.0","canonical_sha256":"e2540feea869d593e79527613ef96f5f8a1e132bfe3a34b9b4bd4c914f374605","source":{"kind":"arxiv","id":"2004.11714","version":1},"attestation_state":"computed","paper":{"title":"Residual Energy-Based Models for Text Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Anton Bakhtin, Arthur Szlam, Marc'Aurelio Ranzato, Myle Ott, Yuntian Deng","submitted_at":"2020-04-22T23:19:55Z","abstract_excerpt":"Text generation is ubiquitous in many NLP tasks, from summarization, to dialogue and machine translation. The dominant parametric approach is based on locally normalized models which predict one word at a time. While these work remarkably well, they are plagued by exposure bias due to the greedy nature of the generation process. In this work, we investigate un-normalized energy-based models (EBMs) which operate not at the token but at the sequence level. In order to make training tractable, we first work in the residual of a pretrained locally normalized language model and second we train 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":"2004.11714","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-22T23:19:55Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"3a86d39f327c10120dcc8a1c5308bcd7239ec8f4edf1822818f0cbfcd361508e","abstract_canon_sha256":"4654d22123e18651be82c7be54a031bff42ce5f528251f8eb8a3f5be3d735e96"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:57:56.919847Z","signature_b64":"Z1zjq6z60UG4+RlUp9Bu9B9wgS2UCUplSazAqiVgwtBHTlHQw7eYUwuTXelNU62v1Wu3RJjEoo01vT9Gm4N2DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e2540feea869d593e79527613ef96f5f8a1e132bfe3a34b9b4bd4c914f374605","last_reissued_at":"2026-07-05T00:57:56.919358Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:57:56.919358Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Residual Energy-Based Models for Text Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Anton Bakhtin, Arthur Szlam, Marc'Aurelio Ranzato, Myle Ott, Yuntian Deng","submitted_at":"2020-04-22T23:19:55Z","abstract_excerpt":"Text generation is ubiquitous in many NLP tasks, from summarization, to dialogue and machine translation. The dominant parametric approach is based on locally normalized models which predict one word at a time. While these work remarkably well, they are plagued by exposure bias due to the greedy nature of the generation process. In this work, we investigate un-normalized energy-based models (EBMs) which operate not at the token but at the sequence level. In order to make training tractable, we first work in the residual of a pretrained locally normalized language model and second we train usin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.11714","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/2004.11714/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":"2004.11714","created_at":"2026-07-05T00:57:56.919426+00:00"},{"alias_kind":"arxiv_version","alias_value":"2004.11714v1","created_at":"2026-07-05T00:57:56.919426+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.11714","created_at":"2026-07-05T00:57:56.919426+00:00"},{"alias_kind":"pith_short_12","alias_value":"4JKA73VINHKZ","created_at":"2026-07-05T00:57:56.919426+00:00"},{"alias_kind":"pith_short_16","alias_value":"4JKA73VINHKZHZ4V","created_at":"2026-07-05T00:57:56.919426+00:00"},{"alias_kind":"pith_short_8","alias_value":"4JKA73VI","created_at":"2026-07-05T00:57:56.919426+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12234","citing_title":"On The Effectiveness-Fluency Trade-Off In LLM Conditioning: A Systematic Study","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2310.06114","citing_title":"Learning Interactive Real-World Simulators","ref_index":218,"is_internal_anchor":false},{"citing_arxiv_id":"2404.12377","citing_title":"RoboDreamer: Learning Compositional World Models for Robot Imagination","ref_index":81,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08645","citing_title":"Energy-based models for diagnostic reconstruction and analysis in a laboratory plasma device","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2006.11239","citing_title":"Denoising Diffusion Probabilistic Models","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6","json":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6.json","graph_json":"https://pith.science/api/pith-number/4JKA73VINHKZHZ4VE5QT56LPL6/graph.json","events_json":"https://pith.science/api/pith-number/4JKA73VINHKZHZ4VE5QT56LPL6/events.json","paper":"https://pith.science/paper/4JKA73VI"},"agent_actions":{"view_html":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6","download_json":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6.json","view_paper":"https://pith.science/paper/4JKA73VI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2004.11714&json=true","fetch_graph":"https://pith.science/api/pith-number/4JKA73VINHKZHZ4VE5QT56LPL6/graph.json","fetch_events":"https://pith.science/api/pith-number/4JKA73VINHKZHZ4VE5QT56LPL6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6/action/storage_attestation","attest_author":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6/action/author_attestation","sign_citation":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6/action/citation_signature","submit_replication":"https://pith.science/pith/4JKA73VINHKZHZ4VE5QT56LPL6/action/replication_record"}},"created_at":"2026-07-05T00:57:56.919426+00:00","updated_at":"2026-07-05T00:57:56.919426+00:00"}