{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:SP4GRREMNAP7D6FPMSALKGY6WK","short_pith_number":"pith:SP4GRREM","schema_version":"1.0","canonical_sha256":"93f868c48c681ff1f8af6480b51b1eb2bec1447e63c2021664d056e83af9a5b0","source":{"kind":"arxiv","id":"1906.05664","version":1},"attestation_state":"computed","paper":{"title":"Calibration, Entropy Rates, and Memory in Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.CL","authors_text":"Cyril Zhang, Karthik Narasimhan, Mark Braverman, Sham M. Kakade, Xinyi Chen, Yi Zhang","submitted_at":"2019-06-11T17:00:49Z","abstract_excerpt":"Building accurate language models that capture meaningful long-term dependencies is a core challenge in natural language processing. Towards this end, we present a calibration-based approach to measure long-term discrepancies between a generative sequence model and the true distribution, and use these discrepancies to improve the model. Empirically, we show that state-of-the-art language models, including LSTMs and Transformers, are \\emph{miscalibrated}: the entropy rates of their generations drift dramatically upward over time. We then provide provable methods to mitigate this phenomenon. Fur"},"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":"1906.05664","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-06-11T17:00:49Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"b849f70bd7e4e29c8ffb06e41a41e38795d8c62c1c43aad298a91030747e7615","abstract_canon_sha256":"838c6b2fdcbf45d076186b16ef66c3fe03069fa0515faf939dd23769491562a7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:43:24.667795Z","signature_b64":"qg3JZyjHR4ye3hZrHEyL0F0wXjPEA+06mMsDKYr55k+9VOSyQ3jlORy0J19DTFbEAbiNVqhEIenSSZHvGvQuDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"93f868c48c681ff1f8af6480b51b1eb2bec1447e63c2021664d056e83af9a5b0","last_reissued_at":"2026-05-17T23:43:24.667380Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:43:24.667380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Calibration, Entropy Rates, and Memory in Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.CL","authors_text":"Cyril Zhang, Karthik Narasimhan, Mark Braverman, Sham M. Kakade, Xinyi Chen, Yi Zhang","submitted_at":"2019-06-11T17:00:49Z","abstract_excerpt":"Building accurate language models that capture meaningful long-term dependencies is a core challenge in natural language processing. Towards this end, we present a calibration-based approach to measure long-term discrepancies between a generative sequence model and the true distribution, and use these discrepancies to improve the model. Empirically, we show that state-of-the-art language models, including LSTMs and Transformers, are \\emph{miscalibrated}: the entropy rates of their generations drift dramatically upward over time. We then provide provable methods to mitigate this phenomenon. Fur"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1906.05664","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":""},"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":"1906.05664","created_at":"2026-05-17T23:43:24.667447+00:00"},{"alias_kind":"arxiv_version","alias_value":"1906.05664v1","created_at":"2026-05-17T23:43:24.667447+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.05664","created_at":"2026-05-17T23:43:24.667447+00:00"},{"alias_kind":"pith_short_12","alias_value":"SP4GRREMNAP7","created_at":"2026-05-18T12:33:27.125529+00:00"},{"alias_kind":"pith_short_16","alias_value":"SP4GRREMNAP7D6FP","created_at":"2026-05-18T12:33:27.125529+00:00"},{"alias_kind":"pith_short_8","alias_value":"SP4GRREM","created_at":"2026-05-18T12:33:27.125529+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK","json":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK.json","graph_json":"https://pith.science/api/pith-number/SP4GRREMNAP7D6FPMSALKGY6WK/graph.json","events_json":"https://pith.science/api/pith-number/SP4GRREMNAP7D6FPMSALKGY6WK/events.json","paper":"https://pith.science/paper/SP4GRREM"},"agent_actions":{"view_html":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK","download_json":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK.json","view_paper":"https://pith.science/paper/SP4GRREM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1906.05664&json=true","fetch_graph":"https://pith.science/api/pith-number/SP4GRREMNAP7D6FPMSALKGY6WK/graph.json","fetch_events":"https://pith.science/api/pith-number/SP4GRREMNAP7D6FPMSALKGY6WK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/action/storage_attestation","attest_author":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/action/author_attestation","sign_citation":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/action/citation_signature","submit_replication":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/action/replication_record"}},"created_at":"2026-05-17T23:43:24.667447+00:00","updated_at":"2026-05-17T23:43:24.667447+00:00"}