{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2017:UIGP5L44YGMZM7STHHQFZ25BFX","short_pith_number":"pith:UIGP5L44","schema_version":"1.0","canonical_sha256":"a20cfeaf9cc199967e5339e05ceba12dfd9ba098b1f9721281ad387aac742543","source":{"kind":"arxiv","id":"1709.07432","version":2},"attestation_state":"computed","paper":{"title":"Dynamic Evaluation of Neural Sequence Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.NE","authors_text":"Ben Krause, Emmanuel Kahembwe, Iain Murray, Steve Renals","submitted_at":"2017-09-21T17:50:04Z","abstract_excerpt":"We present methodology for using dynamic evaluation to improve neural sequence models. Models are adapted to recent history via a gradient descent based mechanism, causing them to assign higher probabilities to re-occurring sequential patterns. Dynamic evaluation outperforms existing adaptation approaches in our comparisons. Dynamic evaluation improves the state-of-the-art word-level perplexities on the Penn Treebank and WikiText-2 datasets to 51.1 and 44.3 respectively, and the state-of-the-art character-level cross-entropies on the text8 and Hutter Prize datasets to 1.19 bits/char and 1.08 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":"1709.07432","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.NE","submitted_at":"2017-09-21T17:50:04Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"9aad1028a8422117597e55dc13d679efe4f06de742921bc09a2e43d3e42f7ff2","abstract_canon_sha256":"9bf6e9d03fbaee71cea9e5d22e9cb6d475a8908b915990f17b275a0e7e5f2f95"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:31:59.505462Z","signature_b64":"aJvh/iBYrOgQ6J63LYxf+YWh71+hMvvcc6yfkl42BK8e3e8kGWHD46JSiaevFujIX4t9gncKYoR2EckNVoztDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a20cfeaf9cc199967e5339e05ceba12dfd9ba098b1f9721281ad387aac742543","last_reissued_at":"2026-05-18T00:31:59.505039Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:31:59.505039Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dynamic Evaluation of Neural Sequence Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.NE","authors_text":"Ben Krause, Emmanuel Kahembwe, Iain Murray, Steve Renals","submitted_at":"2017-09-21T17:50:04Z","abstract_excerpt":"We present methodology for using dynamic evaluation to improve neural sequence models. Models are adapted to recent history via a gradient descent based mechanism, causing them to assign higher probabilities to re-occurring sequential patterns. Dynamic evaluation outperforms existing adaptation approaches in our comparisons. Dynamic evaluation improves the state-of-the-art word-level perplexities on the Penn Treebank and WikiText-2 datasets to 51.1 and 44.3 respectively, and the state-of-the-art character-level cross-entropies on the text8 and Hutter Prize datasets to 1.19 bits/char and 1.08 b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1709.07432","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":""},"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":"1709.07432","created_at":"2026-05-18T00:31:59.505091+00:00"},{"alias_kind":"arxiv_version","alias_value":"1709.07432v2","created_at":"2026-05-18T00:31:59.505091+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1709.07432","created_at":"2026-05-18T00:31:59.505091+00:00"},{"alias_kind":"pith_short_12","alias_value":"UIGP5L44YGMZ","created_at":"2026-05-18T12:31:46.661854+00:00"},{"alias_kind":"pith_short_16","alias_value":"UIGP5L44YGMZM7ST","created_at":"2026-05-18T12:31:46.661854+00:00"},{"alias_kind":"pith_short_8","alias_value":"UIGP5L44","created_at":"2026-05-18T12:31:46.661854+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"1907.00455","citing_title":"Multiplicative Models for Recurrent Language Modeling","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2604.13546","citing_title":"Learning Inference Concurrency in DynamicGate MLP Structural and Mathematical Justification","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX","json":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX.json","graph_json":"https://pith.science/api/pith-number/UIGP5L44YGMZM7STHHQFZ25BFX/graph.json","events_json":"https://pith.science/api/pith-number/UIGP5L44YGMZM7STHHQFZ25BFX/events.json","paper":"https://pith.science/paper/UIGP5L44"},"agent_actions":{"view_html":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX","download_json":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX.json","view_paper":"https://pith.science/paper/UIGP5L44","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1709.07432&json=true","fetch_graph":"https://pith.science/api/pith-number/UIGP5L44YGMZM7STHHQFZ25BFX/graph.json","fetch_events":"https://pith.science/api/pith-number/UIGP5L44YGMZM7STHHQFZ25BFX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX/action/storage_attestation","attest_author":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX/action/author_attestation","sign_citation":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX/action/citation_signature","submit_replication":"https://pith.science/pith/UIGP5L44YGMZM7STHHQFZ25BFX/action/replication_record"}},"created_at":"2026-05-18T00:31:59.505091+00:00","updated_at":"2026-05-18T00:31:59.505091+00:00"}