{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KBUF7ZEOR7I7V35MC4HMUTUGCE","short_pith_number":"pith:KBUF7ZEO","schema_version":"1.0","canonical_sha256":"50685fe48e8fd1faefac170eca4e861119f16d1203e9ef41269ff7e18c93f476","source":{"kind":"arxiv","id":"2509.18085","version":4},"attestation_state":"computed","paper":{"title":"Structuring The Future: Diffusion LLM Speculative Decoding via Calibrated Draft Graphs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Christopher Lott, Fatih Porikli, Mingu Lee, Raghavv Goel, Risheek Garrepalli, Sudhanshu Agrawal","submitted_at":"2025-09-22T17:58:21Z","abstract_excerpt":"Diffusion LLMs (dLLMs) have recently emerged as a powerful alternative to autoregressive LLMs (AR-LLMs) with the potential to operate at significantly higher token-generation rates. To unlock this potential, we present Spiffy, a speculative decoding algorithm to accelerate dLLM inference while provably preserving the model's output distribution. This work addresses the unique challenges involved in applying ideas from speculative decoding of AR-LLMs to dLLMs. Spiffy performs auto-speculation to eliminate the overheads of an independent draft model, structuring draft states in the form of a nov"},"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":"2509.18085","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-22T17:58:21Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"4dad3ca49ef271e730eb017c3aebbbdfbe98bb5c22962daeac8fb94b647c20ec","abstract_canon_sha256":"1f86b7bffeb0994c9b4fbcbe4820d9d43c5dc6107942395f276621ece1f807bd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-12T01:08:16.135155Z","signature_b64":"J1PENtiENd5WJuRc48zASgm5SK0zha8W0ENhkuYcKvi1bqG29z8u0IXtHqzzEZV3fmDp3Zaa7+cTOheMT02JAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"50685fe48e8fd1faefac170eca4e861119f16d1203e9ef41269ff7e18c93f476","last_reissued_at":"2026-06-12T01:08:16.134113Z","signature_status":"signed_v1","first_computed_at":"2026-06-12T01:08:16.134113Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Structuring The Future: Diffusion LLM Speculative Decoding via Calibrated Draft Graphs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Christopher Lott, Fatih Porikli, Mingu Lee, Raghavv Goel, Risheek Garrepalli, Sudhanshu Agrawal","submitted_at":"2025-09-22T17:58:21Z","abstract_excerpt":"Diffusion LLMs (dLLMs) have recently emerged as a powerful alternative to autoregressive LLMs (AR-LLMs) with the potential to operate at significantly higher token-generation rates. To unlock this potential, we present Spiffy, a speculative decoding algorithm to accelerate dLLM inference while provably preserving the model's output distribution. This work addresses the unique challenges involved in applying ideas from speculative decoding of AR-LLMs to dLLMs. Spiffy performs auto-speculation to eliminate the overheads of an independent draft model, structuring draft states in the form of a nov"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.18085","kind":"arxiv","version":4},"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/2509.18085/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":"2509.18085","created_at":"2026-06-12T01:08:16.134259+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.18085v4","created_at":"2026-06-12T01:08:16.134259+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.18085","created_at":"2026-06-12T01:08:16.134259+00:00"},{"alias_kind":"pith_short_12","alias_value":"KBUF7ZEOR7I7","created_at":"2026-06-12T01:08:16.134259+00:00"},{"alias_kind":"pith_short_16","alias_value":"KBUF7ZEOR7I7V35M","created_at":"2026-06-12T01:08:16.134259+00:00"},{"alias_kind":"pith_short_8","alias_value":"KBUF7ZEO","created_at":"2026-06-12T01:08:16.134259+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2602.08404","citing_title":"TEAM: Temporal-Spatial Consistency Guided Expert Activation for MoE Diffusion Language Model Acceleration","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18817","citing_title":"Multi-Token Residual Prediction","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15609","citing_title":"PSD: Pushing the Pareto Frontier of Diffusion LLMs via Parallel Speculative Decoding","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2508.19982","citing_title":"Diffusion Language Models Know the Answer Before Decoding","ref_index":1,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE","json":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE.json","graph_json":"https://pith.science/api/pith-number/KBUF7ZEOR7I7V35MC4HMUTUGCE/graph.json","events_json":"https://pith.science/api/pith-number/KBUF7ZEOR7I7V35MC4HMUTUGCE/events.json","paper":"https://pith.science/paper/KBUF7ZEO"},"agent_actions":{"view_html":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE","download_json":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE.json","view_paper":"https://pith.science/paper/KBUF7ZEO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.18085&json=true","fetch_graph":"https://pith.science/api/pith-number/KBUF7ZEOR7I7V35MC4HMUTUGCE/graph.json","fetch_events":"https://pith.science/api/pith-number/KBUF7ZEOR7I7V35MC4HMUTUGCE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE/action/storage_attestation","attest_author":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE/action/author_attestation","sign_citation":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE/action/citation_signature","submit_replication":"https://pith.science/pith/KBUF7ZEOR7I7V35MC4HMUTUGCE/action/replication_record"}},"created_at":"2026-06-12T01:08:16.134259+00:00","updated_at":"2026-06-12T01:08:16.134259+00:00"}