{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:53SQAMVSTF2VJW32VXTLWJLKLL","short_pith_number":"pith:53SQAMVS","canonical_record":{"source":{"id":"2502.18779","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-26T03:22:44Z","cross_cats_sorted":["cs.DS"],"title_canon_sha256":"e245779392c72fb0d2d8afbe6c60611461ebdc2bc5abb26fd1a9abd16312c8b9","abstract_canon_sha256":"2605238353f83cc96f607c203afa5a4a2aff3e02a672468c27f7769b6fd4f4af"},"schema_version":"1.0"},"canonical_sha256":"eee50032b2997554db7aade6bb256a5ac8bf9d51369d8da38bb1761b73a6c79d","source":{"kind":"arxiv","id":"2502.18779","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.18779","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"arxiv_version","alias_value":"2502.18779v1","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.18779","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"pith_short_12","alias_value":"53SQAMVSTF2V","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"pith_short_16","alias_value":"53SQAMVSTF2VJW32","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"pith_short_8","alias_value":"53SQAMVS","created_at":"2026-07-05T10:20:21Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:53SQAMVSTF2VJW32VXTLWJLKLL","target":"record","payload":{"canonical_record":{"source":{"id":"2502.18779","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-26T03:22:44Z","cross_cats_sorted":["cs.DS"],"title_canon_sha256":"e245779392c72fb0d2d8afbe6c60611461ebdc2bc5abb26fd1a9abd16312c8b9","abstract_canon_sha256":"2605238353f83cc96f607c203afa5a4a2aff3e02a672468c27f7769b6fd4f4af"},"schema_version":"1.0"},"canonical_sha256":"eee50032b2997554db7aade6bb256a5ac8bf9d51369d8da38bb1761b73a6c79d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:20:21.363235Z","signature_b64":"G8mBBtTi1VAKwLIQYDRmhKFtrRRWH0BxOwvjiOepWv2ULyLTLzk/FGTE6H4zacGhk4jcrx5XgL9kqiWNYEfGCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eee50032b2997554db7aade6bb256a5ac8bf9d51369d8da38bb1761b73a6c79d","last_reissued_at":"2026-07-05T10:20:21.362708Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:20:21.362708Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.18779","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:20:21Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RoKoUV4Cg+XTzLArAfv2t+LaidliH0g4pZTOkEYHG85tYYU83sUp9KC7nxT4p8GbTKzLNrryfKr0dXlx0xolCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T06:14:25.710475Z"},"content_sha256":"2ce0b538eed73fffd5a36285581016e98584b15886f186f6c4882e1da93f5d54","schema_version":"1.0","event_id":"sha256:2ce0b538eed73fffd5a36285581016e98584b15886f186f6c4882e1da93f5d54"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:53SQAMVSTF2VJW32VXTLWJLKLL","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Towards Optimal Multi-draft Speculative Decoding","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.DS"],"primary_cat":"cs.CL","authors_text":"Dinesh Manocha, Heng Huang, Ryan A. Rossi, Tong Zheng, Vignesh Viswanathan, Yihan Wu, Zhengmian Hu, Ziyi Chen","submitted_at":"2025-02-26T03:22:44Z","abstract_excerpt":"Large Language Models (LLMs) have become an indispensable part of natural language processing tasks. However, autoregressive sampling has become an efficiency bottleneck. Multi-Draft Speculative Decoding (MDSD) is a recent approach where, when generating each token, a small draft model generates multiple drafts, and the target LLM verifies them in parallel, ensuring that the final output conforms to the target model distribution. The two main design choices in MDSD are the draft sampling method and the verification algorithm. For a fixed draft sampling method, the optimal acceptance rate is a "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.18779","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/2502.18779/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:20:21Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"w7/VIcQOEruS1SqR3y44PVgpFnZtVv1DgcBfI5+1K6ydsgItC8O/ltPaeZFfTwEq7oHTly2W7gIUMHrtm4GKBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T06:14:25.710851Z"},"content_sha256":"006e10063439d31f5478317e59eb4c6163d164d0bc84ec3650c0858c7f5c3abf","schema_version":"1.0","event_id":"sha256:006e10063439d31f5478317e59eb4c6163d164d0bc84ec3650c0858c7f5c3abf"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/53SQAMVSTF2VJW32VXTLWJLKLL/bundle.json","state_url":"https://pith.science/pith/53SQAMVSTF2VJW32VXTLWJLKLL/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/53SQAMVSTF2VJW32VXTLWJLKLL/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-07-07T06:14:25Z","links":{"resolver":"https://pith.science/pith/53SQAMVSTF2VJW32VXTLWJLKLL","bundle":"https://pith.science/pith/53SQAMVSTF2VJW32VXTLWJLKLL/bundle.json","state":"https://pith.science/pith/53SQAMVSTF2VJW32VXTLWJLKLL/state.json","well_known_bundle":"https://pith.science/.well-known/pith/53SQAMVSTF2VJW32VXTLWJLKLL/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:53SQAMVSTF2VJW32VXTLWJLKLL","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"2605238353f83cc96f607c203afa5a4a2aff3e02a672468c27f7769b6fd4f4af","cross_cats_sorted":["cs.DS"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-26T03:22:44Z","title_canon_sha256":"e245779392c72fb0d2d8afbe6c60611461ebdc2bc5abb26fd1a9abd16312c8b9"},"schema_version":"1.0","source":{"id":"2502.18779","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.18779","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"arxiv_version","alias_value":"2502.18779v1","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.18779","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"pith_short_12","alias_value":"53SQAMVSTF2V","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"pith_short_16","alias_value":"53SQAMVSTF2VJW32","created_at":"2026-07-05T10:20:21Z"},{"alias_kind":"pith_short_8","alias_value":"53SQAMVS","created_at":"2026-07-05T10:20:21Z"}],"graph_snapshots":[{"event_id":"sha256:006e10063439d31f5478317e59eb4c6163d164d0bc84ec3650c0858c7f5c3abf","target":"graph","created_at":"2026-07-05T10:20:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2502.18779/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) have become an indispensable part of natural language processing tasks. However, autoregressive sampling has become an efficiency bottleneck. Multi-Draft Speculative Decoding (MDSD) is a recent approach where, when generating each token, a small draft model generates multiple drafts, and the target LLM verifies them in parallel, ensuring that the final output conforms to the target model distribution. The two main design choices in MDSD are the draft sampling method and the verification algorithm. For a fixed draft sampling method, the optimal acceptance rate is a ","authors_text":"Dinesh Manocha, Heng Huang, Ryan A. Rossi, Tong Zheng, Vignesh Viswanathan, Yihan Wu, Zhengmian Hu, Ziyi Chen","cross_cats":["cs.DS"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-26T03:22:44Z","title":"Towards Optimal Multi-draft Speculative Decoding"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.18779","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:2ce0b538eed73fffd5a36285581016e98584b15886f186f6c4882e1da93f5d54","target":"record","created_at":"2026-07-05T10:20:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"2605238353f83cc96f607c203afa5a4a2aff3e02a672468c27f7769b6fd4f4af","cross_cats_sorted":["cs.DS"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-26T03:22:44Z","title_canon_sha256":"e245779392c72fb0d2d8afbe6c60611461ebdc2bc5abb26fd1a9abd16312c8b9"},"schema_version":"1.0","source":{"id":"2502.18779","kind":"arxiv","version":1}},"canonical_sha256":"eee50032b2997554db7aade6bb256a5ac8bf9d51369d8da38bb1761b73a6c79d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"eee50032b2997554db7aade6bb256a5ac8bf9d51369d8da38bb1761b73a6c79d","first_computed_at":"2026-07-05T10:20:21.362708Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:20:21.362708Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"G8mBBtTi1VAKwLIQYDRmhKFtrRRWH0BxOwvjiOepWv2ULyLTLzk/FGTE6H4zacGhk4jcrx5XgL9kqiWNYEfGCQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:20:21.363235Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.18779","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:2ce0b538eed73fffd5a36285581016e98584b15886f186f6c4882e1da93f5d54","sha256:006e10063439d31f5478317e59eb4c6163d164d0bc84ec3650c0858c7f5c3abf"],"state_sha256":"666a6fe5d4353adf902ff9e6d323f985b74edcb0bc7c8de66d651137b55aa565"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/DPLesEXalUyiUfnNajY1fjqxQj0X3zK9kKvG0uZcpzo9Oqoqf8wzLfVFdNDN2rY+D/KBJAwL1/hHjcyU4fXDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T06:14:25.712872Z","bundle_sha256":"c7407e74f528b0393169f4ced32b542e00199b26faaf095008c4f7504bfb46e3"}}