{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:F66YKWTR5QMIZP3ZMTXSRSHKOT","short_pith_number":"pith:F66YKWTR","schema_version":"1.0","canonical_sha256":"2fbd855a71ec188cbf7964ef28c8ea74db601b1a208bbd80e7585c77964673d9","source":{"kind":"arxiv","id":"1906.02735","version":6},"attestation_state":"computed","paper":{"title":"Residual Flows for Invertible Generative Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"David Duvenaud, Jens Behrmann, J\\\"orn-Henrik Jacobsen, Ricky T. Q. Chen","submitted_at":"2019-06-06T17:55:01Z","abstract_excerpt":"Flow-based generative models parameterize probability distributions through an invertible transformation and can be trained by maximum likelihood. Invertible residual networks provide a flexible family of transformations where only Lipschitz conditions rather than strict architectural constraints are needed for enforcing invertibility. However, prior work trained invertible residual networks for density estimation by relying on biased log-density estimates whose bias increased with the network's expressiveness. We give a tractable unbiased estimate of the log density using a \"Russian roulette\""},"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":"1906.02735","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2019-06-06T17:55:01Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"2593749b009c558b1e3568d1911c55a66ab9686829c244deda21b1240ea591e0","abstract_canon_sha256":"58d3241a51a2e8e3296abd1453380f1061958ff5cd3511ad7d6bed353cbb4be6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:21:45.894226Z","signature_b64":"h90B3lalHRwskwnneUIN2YK8QW7v2zax5ok4v/k+C6BUk2t7oWmVgJ9c7GmBxzX+BcxJsu+xBCtAxRDMzwClDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2fbd855a71ec188cbf7964ef28c8ea74db601b1a208bbd80e7585c77964673d9","last_reissued_at":"2026-07-05T01:21:45.893725Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:21:45.893725Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Residual Flows for Invertible Generative Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"David Duvenaud, Jens Behrmann, J\\\"orn-Henrik Jacobsen, Ricky T. Q. Chen","submitted_at":"2019-06-06T17:55:01Z","abstract_excerpt":"Flow-based generative models parameterize probability distributions through an invertible transformation and can be trained by maximum likelihood. Invertible residual networks provide a flexible family of transformations where only Lipschitz conditions rather than strict architectural constraints are needed for enforcing invertibility. However, prior work trained invertible residual networks for density estimation by relying on biased log-density estimates whose bias increased with the network's expressiveness. We give a tractable unbiased estimate of the log density using a \"Russian roulette\""},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1906.02735","kind":"arxiv","version":6},"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/1906.02735/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":"1906.02735","created_at":"2026-07-05T01:21:45.893785+00:00"},{"alias_kind":"arxiv_version","alias_value":"1906.02735v6","created_at":"2026-07-05T01:21:45.893785+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.02735","created_at":"2026-07-05T01:21:45.893785+00:00"},{"alias_kind":"pith_short_12","alias_value":"F66YKWTR5QMI","created_at":"2026-07-05T01:21:45.893785+00:00"},{"alias_kind":"pith_short_16","alias_value":"F66YKWTR5QMIZP3Z","created_at":"2026-07-05T01:21:45.893785+00:00"},{"alias_kind":"pith_short_8","alias_value":"F66YKWTR","created_at":"2026-07-05T01:21:45.893785+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.22444","citing_title":"Normalizing flows for all-orders QED corrections in lattice field theory","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2308.08089","citing_title":"DragNUWA: Fine-grained Control in Video Generation by Integrating Text, Image, and Trajectory","ref_index":144,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT","json":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT.json","graph_json":"https://pith.science/api/pith-number/F66YKWTR5QMIZP3ZMTXSRSHKOT/graph.json","events_json":"https://pith.science/api/pith-number/F66YKWTR5QMIZP3ZMTXSRSHKOT/events.json","paper":"https://pith.science/paper/F66YKWTR"},"agent_actions":{"view_html":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT","download_json":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT.json","view_paper":"https://pith.science/paper/F66YKWTR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1906.02735&json=true","fetch_graph":"https://pith.science/api/pith-number/F66YKWTR5QMIZP3ZMTXSRSHKOT/graph.json","fetch_events":"https://pith.science/api/pith-number/F66YKWTR5QMIZP3ZMTXSRSHKOT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT/action/storage_attestation","attest_author":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT/action/author_attestation","sign_citation":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT/action/citation_signature","submit_replication":"https://pith.science/pith/F66YKWTR5QMIZP3ZMTXSRSHKOT/action/replication_record"}},"created_at":"2026-07-05T01:21:45.893785+00:00","updated_at":"2026-07-05T01:21:45.893785+00:00"}