{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:WV33OF2UFEB2UKRGANREGZ6DLC","short_pith_number":"pith:WV33OF2U","schema_version":"1.0","canonical_sha256":"b577b717542903aa2a2603624367c358bc3f4e4c3d9f52994dba964e763e0248","source":{"kind":"arxiv","id":"2006.01659","version":1},"attestation_state":"computed","paper":{"title":"Surprisal-Triggered Conditional Computation with Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Brett H. Meyer, Derek Nowrouzezahrai, Loren Lugosch","submitted_at":"2020-06-02T14:34:24Z","abstract_excerpt":"Autoregressive neural network models have been used successfully for sequence generation, feature extraction, and hypothesis scoring. This paper presents yet another use for these models: allocating more computation to more difficult inputs. In our model, an autoregressive model is used both to extract features and to predict observations in a stream of input observations. The surprisal of the input, measured as the negative log-likelihood of the current observation according to the autoregressive model, is used as a measure of input difficulty. This in turn determines whether a small, fast ne"},"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":"2006.01659","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-02T14:34:24Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"3a0e8ec7fb2a3f2a3f67cca76037b8038b44a152400ef8f851dee3a8bbdcf4b3","abstract_canon_sha256":"7d57944bfe1c3372ce4c7796b21c75e472d22bb4f0977943a396041ec95e87db"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:07:33.092225Z","signature_b64":"WBoM87mXynJwbRqGxzmGienrXpk6TIef8wFmGxvUNce2sjj72Seldqcf3m14Ykt0hYD3TLb8ZVW+F/FsAwiLBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b577b717542903aa2a2603624367c358bc3f4e4c3d9f52994dba964e763e0248","last_reissued_at":"2026-07-05T01:07:33.091847Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:07:33.091847Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Surprisal-Triggered Conditional Computation with Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Brett H. Meyer, Derek Nowrouzezahrai, Loren Lugosch","submitted_at":"2020-06-02T14:34:24Z","abstract_excerpt":"Autoregressive neural network models have been used successfully for sequence generation, feature extraction, and hypothesis scoring. This paper presents yet another use for these models: allocating more computation to more difficult inputs. In our model, an autoregressive model is used both to extract features and to predict observations in a stream of input observations. The surprisal of the input, measured as the negative log-likelihood of the current observation according to the autoregressive model, is used as a measure of input difficulty. This in turn determines whether a small, fast ne"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.01659","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/2006.01659/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":"2006.01659","created_at":"2026-07-05T01:07:33.091910+00:00"},{"alias_kind":"arxiv_version","alias_value":"2006.01659v1","created_at":"2026-07-05T01:07:33.091910+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.01659","created_at":"2026-07-05T01:07:33.091910+00:00"},{"alias_kind":"pith_short_12","alias_value":"WV33OF2UFEB2","created_at":"2026-07-05T01:07:33.091910+00:00"},{"alias_kind":"pith_short_16","alias_value":"WV33OF2UFEB2UKRG","created_at":"2026-07-05T01:07:33.091910+00:00"},{"alias_kind":"pith_short_8","alias_value":"WV33OF2U","created_at":"2026-07-05T01:07:33.091910+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.16421","citing_title":"When to Plan: Learning to Select Between Reactive Control and Deliberative Planning","ref_index":21,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC","json":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC.json","graph_json":"https://pith.science/api/pith-number/WV33OF2UFEB2UKRGANREGZ6DLC/graph.json","events_json":"https://pith.science/api/pith-number/WV33OF2UFEB2UKRGANREGZ6DLC/events.json","paper":"https://pith.science/paper/WV33OF2U"},"agent_actions":{"view_html":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC","download_json":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC.json","view_paper":"https://pith.science/paper/WV33OF2U","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2006.01659&json=true","fetch_graph":"https://pith.science/api/pith-number/WV33OF2UFEB2UKRGANREGZ6DLC/graph.json","fetch_events":"https://pith.science/api/pith-number/WV33OF2UFEB2UKRGANREGZ6DLC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC/action/storage_attestation","attest_author":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC/action/author_attestation","sign_citation":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC/action/citation_signature","submit_replication":"https://pith.science/pith/WV33OF2UFEB2UKRGANREGZ6DLC/action/replication_record"}},"created_at":"2026-07-05T01:07:33.091910+00:00","updated_at":"2026-07-05T01:07:33.091910+00:00"}