{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QOGCOYU7PR2OS65I4YKQPIBKDJ","short_pith_number":"pith:QOGCOYU7","schema_version":"1.0","canonical_sha256":"838c27629f7c74e97ba8e61507a02a1a71e31c86d3d3bd254ecdb3f348ad2584","source":{"kind":"arxiv","id":"2606.27474","version":1},"attestation_state":"computed","paper":{"title":"Speculative Refinement: A Hybrid Autoregressive Diffusion Decoding Strategy and Its Behavior Across Benchmarks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Aditi Gupta, Kushagra Trivedi, Neel Mishra, Pawan Kumar","submitted_at":"2026-06-25T18:52:24Z","abstract_excerpt":"How should we evaluate generation systems that combine autoregressive (AR) and diffusion decoding? We study this question through Speculative Refinement (SpecRef), a training-free hybrid method that warm-starts a masked diffusion language model from an AR draft using entropy-guided selective masking. Evaluating SpecRef across six benchmarks (HumanEval, MBPP, GSM8K, BBH, ARC-Challenge, HellaSwag) with three distinct evaluation protocols (execution-based pass@1, exact-match, log-likelihood scoring), we surface several findings relevant beyond our specific system: (1) code benchmarks conflate str"},"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":"2606.27474","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2026-06-25T18:52:24Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2659084f62a2782caef188e9ea035e55e1445cb88501860f63517c261cfc8cbc","abstract_canon_sha256":"5cf0d6f810bc311d19cc84b65f3fda6e106a1378a514609b432ac2feb27ccf94"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-29T00:14:07.423900Z","signature_b64":"G+/w255UzUM9fazX9JrB3zTzaycfZ5zQHnY44ZzgpZYX+Z4mgAxirF8GEYN5vljvpZl5AC2yr9kuIvj7GdG+BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"838c27629f7c74e97ba8e61507a02a1a71e31c86d3d3bd254ecdb3f348ad2584","last_reissued_at":"2026-06-29T00:14:07.423451Z","signature_status":"signed_v1","first_computed_at":"2026-06-29T00:14:07.423451Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Speculative Refinement: A Hybrid Autoregressive Diffusion Decoding Strategy and Its Behavior Across Benchmarks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Aditi Gupta, Kushagra Trivedi, Neel Mishra, Pawan Kumar","submitted_at":"2026-06-25T18:52:24Z","abstract_excerpt":"How should we evaluate generation systems that combine autoregressive (AR) and diffusion decoding? We study this question through Speculative Refinement (SpecRef), a training-free hybrid method that warm-starts a masked diffusion language model from an AR draft using entropy-guided selective masking. Evaluating SpecRef across six benchmarks (HumanEval, MBPP, GSM8K, BBH, ARC-Challenge, HellaSwag) with three distinct evaluation protocols (execution-based pass@1, exact-match, log-likelihood scoring), we surface several findings relevant beyond our specific system: (1) code benchmarks conflate str"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.27474","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/2606.27474/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":"2606.27474","created_at":"2026-06-29T00:14:07.423534+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.27474v1","created_at":"2026-06-29T00:14:07.423534+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.27474","created_at":"2026-06-29T00:14:07.423534+00:00"},{"alias_kind":"pith_short_12","alias_value":"QOGCOYU7PR2O","created_at":"2026-06-29T00:14:07.423534+00:00"},{"alias_kind":"pith_short_16","alias_value":"QOGCOYU7PR2OS65I","created_at":"2026-06-29T00:14:07.423534+00:00"},{"alias_kind":"pith_short_8","alias_value":"QOGCOYU7","created_at":"2026-06-29T00:14:07.423534+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ","json":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ.json","graph_json":"https://pith.science/api/pith-number/QOGCOYU7PR2OS65I4YKQPIBKDJ/graph.json","events_json":"https://pith.science/api/pith-number/QOGCOYU7PR2OS65I4YKQPIBKDJ/events.json","paper":"https://pith.science/paper/QOGCOYU7"},"agent_actions":{"view_html":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ","download_json":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ.json","view_paper":"https://pith.science/paper/QOGCOYU7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.27474&json=true","fetch_graph":"https://pith.science/api/pith-number/QOGCOYU7PR2OS65I4YKQPIBKDJ/graph.json","fetch_events":"https://pith.science/api/pith-number/QOGCOYU7PR2OS65I4YKQPIBKDJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ/action/storage_attestation","attest_author":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ/action/author_attestation","sign_citation":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ/action/citation_signature","submit_replication":"https://pith.science/pith/QOGCOYU7PR2OS65I4YKQPIBKDJ/action/replication_record"}},"created_at":"2026-06-29T00:14:07.423534+00:00","updated_at":"2026-06-29T00:14:07.423534+00:00"}