{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:SP4GRREMNAP7D6FPMSALKGY6WK","short_pith_number":"pith:SP4GRREM","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"},"canonical_sha256":"93f868c48c681ff1f8af6480b51b1eb2bec1447e63c2021664d056e83af9a5b0","source":{"kind":"arxiv","id":"1906.05664","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1906.05664","created_at":"2026-05-17T23:43:24Z"},{"alias_kind":"arxiv_version","alias_value":"1906.05664v1","created_at":"2026-05-17T23:43:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.05664","created_at":"2026-05-17T23:43:24Z"},{"alias_kind":"pith_short_12","alias_value":"SP4GRREMNAP7","created_at":"2026-05-18T12:33:27Z"},{"alias_kind":"pith_short_16","alias_value":"SP4GRREMNAP7D6FP","created_at":"2026-05-18T12:33:27Z"},{"alias_kind":"pith_short_8","alias_value":"SP4GRREM","created_at":"2026-05-18T12:33:27Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:SP4GRREMNAP7D6FPMSALKGY6WK","target":"record","payload":{"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"},"canonical_sha256":"93f868c48c681ff1f8af6480b51b1eb2bec1447e63c2021664d056e83af9a5b0","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"},"source_kind":"arxiv","source_id":"1906.05664","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:43:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"iKdnpEuAw5ZzQvz5FE0GhGMpkuz5Bm7e1rAtCVTGppG6aZtXDW6kh/gwllBHakoZpLjaSqOb0e23vFd+pJGwAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T11:57:39.513660Z"},"content_sha256":"b7deee63572d05688671a0f55a0279a45eb5dd69f4489a700731eacbfcacea89","schema_version":"1.0","event_id":"sha256:b7deee63572d05688671a0f55a0279a45eb5dd69f4489a700731eacbfcacea89"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:SP4GRREMNAP7D6FPMSALKGY6WK","target":"graph","payload":{"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:43:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HcOCLIlHtsCNVnoMVIgb6VYxOHOf4oDFMrRl1Mki44/UPNsRxY49Go8SEGWbJl6sHJjaPSXOe0i/nqf1cXvTCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T11:57:39.514352Z"},"content_sha256":"83e9670bcaeb5261929286a112664b6dc92b9006d1ba66fc12861bba1c937def","schema_version":"1.0","event_id":"sha256:83e9670bcaeb5261929286a112664b6dc92b9006d1ba66fc12861bba1c937def"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/bundle.json","state_url":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SP4GRREMNAP7D6FPMSALKGY6WK/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-25T11:57:39Z","links":{"resolver":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK","bundle":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/bundle.json","state":"https://pith.science/pith/SP4GRREMNAP7D6FPMSALKGY6WK/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SP4GRREMNAP7D6FPMSALKGY6WK/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:SP4GRREMNAP7D6FPMSALKGY6WK","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"838c6b2fdcbf45d076186b16ef66c3fe03069fa0515faf939dd23769491562a7","cross_cats_sorted":["cs.LG","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-06-11T17:00:49Z","title_canon_sha256":"b849f70bd7e4e29c8ffb06e41a41e38795d8c62c1c43aad298a91030747e7615"},"schema_version":"1.0","source":{"id":"1906.05664","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1906.05664","created_at":"2026-05-17T23:43:24Z"},{"alias_kind":"arxiv_version","alias_value":"1906.05664v1","created_at":"2026-05-17T23:43:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.05664","created_at":"2026-05-17T23:43:24Z"},{"alias_kind":"pith_short_12","alias_value":"SP4GRREMNAP7","created_at":"2026-05-18T12:33:27Z"},{"alias_kind":"pith_short_16","alias_value":"SP4GRREMNAP7D6FP","created_at":"2026-05-18T12:33:27Z"},{"alias_kind":"pith_short_8","alias_value":"SP4GRREM","created_at":"2026-05-18T12:33:27Z"}],"graph_snapshots":[{"event_id":"sha256:83e9670bcaeb5261929286a112664b6dc92b9006d1ba66fc12861bba1c937def","target":"graph","created_at":"2026-05-17T23:43:24Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"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","authors_text":"Cyril Zhang, Karthik Narasimhan, Mark Braverman, Sham M. Kakade, Xinyi Chen, Yi Zhang","cross_cats":["cs.LG","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-06-11T17:00:49Z","title":"Calibration, Entropy Rates, and Memory in Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1906.05664","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:b7deee63572d05688671a0f55a0279a45eb5dd69f4489a700731eacbfcacea89","target":"record","created_at":"2026-05-17T23:43:24Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"838c6b2fdcbf45d076186b16ef66c3fe03069fa0515faf939dd23769491562a7","cross_cats_sorted":["cs.LG","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-06-11T17:00:49Z","title_canon_sha256":"b849f70bd7e4e29c8ffb06e41a41e38795d8c62c1c43aad298a91030747e7615"},"schema_version":"1.0","source":{"id":"1906.05664","kind":"arxiv","version":1}},"canonical_sha256":"93f868c48c681ff1f8af6480b51b1eb2bec1447e63c2021664d056e83af9a5b0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"93f868c48c681ff1f8af6480b51b1eb2bec1447e63c2021664d056e83af9a5b0","first_computed_at":"2026-05-17T23:43:24.667380Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:43:24.667380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"qg3JZyjHR4ye3hZrHEyL0F0wXjPEA+06mMsDKYr55k+9VOSyQ3jlORy0J19DTFbEAbiNVqhEIenSSZHvGvQuDQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:43:24.667795Z","signed_message":"canonical_sha256_bytes"},"source_id":"1906.05664","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b7deee63572d05688671a0f55a0279a45eb5dd69f4489a700731eacbfcacea89","sha256:83e9670bcaeb5261929286a112664b6dc92b9006d1ba66fc12861bba1c937def"],"state_sha256":"b8037a5707d3a1f1d2b94317eae9925f6a9a2abff95cb9f655d1f553df4b5eeb"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4bSgnktRAYAF0lOHNFPvtRJWG8xFh5SP6LX2i7twOErhbKsR43Ls1wEzVsHsfNRKQcVVIrVw050cN8SBfapYDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-25T11:57:39.517770Z","bundle_sha256":"788727f94cca8ccfbd5c4c5e4b45eb0b89ab64cc2f23f2ae70c2fb9a3a81414e"}}