{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:ISOJZDKWLDE6YWZ24QR6E75DO4","short_pith_number":"pith:ISOJZDKW","schema_version":"1.0","canonical_sha256":"449c9c8d5658c9ec5b3ae423e27fa3772761c0b739c5ad7419ac80468dcfc1c8","source":{"kind":"arxiv","id":"1903.06059","version":2},"attestation_state":"computed","paper":{"title":"Stochastic Beams and Where to Find Them: The Gumbel-Top-k Trick for Sampling Sequences Without Replacement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Herke van Hoof, Max Welling, Wouter Kool","submitted_at":"2019-03-14T14:56:06Z","abstract_excerpt":"The well-known Gumbel-Max trick for sampling from a categorical distribution can be extended to sample $k$ elements without replacement. We show how to implicitly apply this 'Gumbel-Top-$k$' trick on a factorized distribution over sequences, allowing to draw exact samples without replacement using a Stochastic Beam Search. Even for exponentially large domains, the number of model evaluations grows only linear in $k$ and the maximum sampled sequence length. The algorithm creates a theoretical connection between sampling and (deterministic) beam search and can be used as a principled intermediat"},"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":"1903.06059","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-03-14T14:56:06Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"d1dcd317f661a87432ff05b9200e2d662fb50d9a52fbe61151fc3759c0abf706","abstract_canon_sha256":"2edb653cec54c0495b37a409e2d4b9d1de7fb7e508ed49e812a63f54f956dae0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:44:42.392849Z","signature_b64":"HukU3mBUIcDWpw3Ye7Q8jDR45GmG6SlYVrAJTx/pDy2ja1bHCn7BgI84saESJ3zeg76xptjq2nUKXRbVsDp6Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"449c9c8d5658c9ec5b3ae423e27fa3772761c0b739c5ad7419ac80468dcfc1c8","last_reissued_at":"2026-05-17T23:44:42.392187Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:44:42.392187Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stochastic Beams and Where to Find Them: The Gumbel-Top-k Trick for Sampling Sequences Without Replacement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Herke van Hoof, Max Welling, Wouter Kool","submitted_at":"2019-03-14T14:56:06Z","abstract_excerpt":"The well-known Gumbel-Max trick for sampling from a categorical distribution can be extended to sample $k$ elements without replacement. We show how to implicitly apply this 'Gumbel-Top-$k$' trick on a factorized distribution over sequences, allowing to draw exact samples without replacement using a Stochastic Beam Search. Even for exponentially large domains, the number of model evaluations grows only linear in $k$ and the maximum sampled sequence length. The algorithm creates a theoretical connection between sampling and (deterministic) beam search and can be used as a principled intermediat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1903.06059","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":"1903.06059","created_at":"2026-05-17T23:44:42.392286+00:00"},{"alias_kind":"arxiv_version","alias_value":"1903.06059v2","created_at":"2026-05-17T23:44:42.392286+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1903.06059","created_at":"2026-05-17T23:44:42.392286+00:00"},{"alias_kind":"pith_short_12","alias_value":"ISOJZDKWLDE6","created_at":"2026-05-18T12:33:18.533446+00:00"},{"alias_kind":"pith_short_16","alias_value":"ISOJZDKWLDE6YWZ2","created_at":"2026-05-18T12:33:18.533446+00:00"},{"alias_kind":"pith_short_8","alias_value":"ISOJZDKW","created_at":"2026-05-18T12:33:18.533446+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.20500","citing_title":"Efficient Test-Time Inference via Deterministic Exploration of Truncated Decoding Trees","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4","json":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4.json","graph_json":"https://pith.science/api/pith-number/ISOJZDKWLDE6YWZ24QR6E75DO4/graph.json","events_json":"https://pith.science/api/pith-number/ISOJZDKWLDE6YWZ24QR6E75DO4/events.json","paper":"https://pith.science/paper/ISOJZDKW"},"agent_actions":{"view_html":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4","download_json":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4.json","view_paper":"https://pith.science/paper/ISOJZDKW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1903.06059&json=true","fetch_graph":"https://pith.science/api/pith-number/ISOJZDKWLDE6YWZ24QR6E75DO4/graph.json","fetch_events":"https://pith.science/api/pith-number/ISOJZDKWLDE6YWZ24QR6E75DO4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4/action/storage_attestation","attest_author":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4/action/author_attestation","sign_citation":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4/action/citation_signature","submit_replication":"https://pith.science/pith/ISOJZDKWLDE6YWZ24QR6E75DO4/action/replication_record"}},"created_at":"2026-05-17T23:44:42.392286+00:00","updated_at":"2026-05-17T23:44:42.392286+00:00"}