{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:UWRSOLDMYYMEZHG3LCTBXEELJO","short_pith_number":"pith:UWRSOLDM","canonical_record":{"source":{"id":"2502.05202","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-31T19:13:58Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"2afeb64d2061d341b56798eae87af9cde5c7a6274cb492640e297b493fd9dbdc","abstract_canon_sha256":"8360f2ab6ab3ed597804e6789d7caff78cd6f2c2c11067b571f156cddede961c"},"schema_version":"1.0"},"canonical_sha256":"a5a3272c6cc6184c9cdb58a61b908b4b88aa0d18600c7f2f3f3654ee563db317","source":{"kind":"arxiv","id":"2502.05202","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.05202","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"arxiv_version","alias_value":"2502.05202v3","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.05202","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"pith_short_12","alias_value":"UWRSOLDMYYME","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"pith_short_16","alias_value":"UWRSOLDMYYMEZHG3","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"pith_short_8","alias_value":"UWRSOLDM","created_at":"2026-07-05T11:19:24Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:UWRSOLDMYYMEZHG3LCTBXEELJO","target":"record","payload":{"canonical_record":{"source":{"id":"2502.05202","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-31T19:13:58Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"2afeb64d2061d341b56798eae87af9cde5c7a6274cb492640e297b493fd9dbdc","abstract_canon_sha256":"8360f2ab6ab3ed597804e6789d7caff78cd6f2c2c11067b571f156cddede961c"},"schema_version":"1.0"},"canonical_sha256":"a5a3272c6cc6184c9cdb58a61b908b4b88aa0d18600c7f2f3f3654ee563db317","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:19:24.454403Z","signature_b64":"87pSf4CR0HCP86t24cJaTs8UcoJ2JXYxT1xQHxJMqS/r1Za+NkuPhjoKU53zCnXS8SydQk7CWd98bqdRFANQCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a5a3272c6cc6184c9cdb58a61b908b4b88aa0d18600c7f2f3f3654ee563db317","last_reissued_at":"2026-07-05T11:19:24.453899Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:19:24.453899Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.05202","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:19:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wDINt+R7ALebNu7mHJzBKTbzrS4FhkqSCC2fv0wRjDlVUziA5Skj51drSnVUKjHAArL+KtsJSvPry2HySGNdDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T02:45:03.902609Z"},"content_sha256":"bc8d93496d4dcb0a30d7d299ce0fc018e75da3d39d694cf3c854874e70075a2e","schema_version":"1.0","event_id":"sha256:bc8d93496d4dcb0a30d7d299ce0fc018e75da3d39d694cf3c854874e70075a2e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:UWRSOLDMYYMEZHG3LCTBXEELJO","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Accelerating LLM Inference with Lossless Speculative Decoding Algorithms for Heterogeneous Vocabularies","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Daniel Korat, David Harel, Gaurav Jain, Jonathan Mamou, Moshe Berchansky, Moshe Wasserblat, Nadav Timor, Oren Pereg","submitted_at":"2025-01-31T19:13:58Z","abstract_excerpt":"Accelerating the inference of large language models (LLMs) is a critical challenge in generative AI. Speculative decoding (SD) methods offer substantial efficiency gains by generating multiple tokens using a single target forward pass. However, existing SD approaches require the drafter and target models to share the same vocabulary, thus limiting the pool of possible drafters, often necessitating the training of a drafter from scratch. We present three new SD methods that remove this shared-vocabulary constraint. All three methods preserve the target distribution (i.e., they are lossless) and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.05202","kind":"arxiv","version":3},"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/2502.05202/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":"2026-07-05T11:19:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"EJ209Wpa/TTewjQGT8gAak2YZTbFp4fC9xZYuUj673I023CBGrTE+vQrJ7zOy1I7N57/QLNj57lugDR39OMLCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T02:45:03.902980Z"},"content_sha256":"537c1cbf3b1c1a4033b442865e4f8fece3828d217edde06466bda28de7ef877b","schema_version":"1.0","event_id":"sha256:537c1cbf3b1c1a4033b442865e4f8fece3828d217edde06466bda28de7ef877b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/UWRSOLDMYYMEZHG3LCTBXEELJO/bundle.json","state_url":"https://pith.science/pith/UWRSOLDMYYMEZHG3LCTBXEELJO/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/UWRSOLDMYYMEZHG3LCTBXEELJO/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":"2026-07-07T02:45:03Z","links":{"resolver":"https://pith.science/pith/UWRSOLDMYYMEZHG3LCTBXEELJO","bundle":"https://pith.science/pith/UWRSOLDMYYMEZHG3LCTBXEELJO/bundle.json","state":"https://pith.science/pith/UWRSOLDMYYMEZHG3LCTBXEELJO/state.json","well_known_bundle":"https://pith.science/.well-known/pith/UWRSOLDMYYMEZHG3LCTBXEELJO/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:UWRSOLDMYYMEZHG3LCTBXEELJO","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":"8360f2ab6ab3ed597804e6789d7caff78cd6f2c2c11067b571f156cddede961c","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-31T19:13:58Z","title_canon_sha256":"2afeb64d2061d341b56798eae87af9cde5c7a6274cb492640e297b493fd9dbdc"},"schema_version":"1.0","source":{"id":"2502.05202","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.05202","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"arxiv_version","alias_value":"2502.05202v3","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.05202","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"pith_short_12","alias_value":"UWRSOLDMYYME","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"pith_short_16","alias_value":"UWRSOLDMYYMEZHG3","created_at":"2026-07-05T11:19:24Z"},{"alias_kind":"pith_short_8","alias_value":"UWRSOLDM","created_at":"2026-07-05T11:19:24Z"}],"graph_snapshots":[{"event_id":"sha256:537c1cbf3b1c1a4033b442865e4f8fece3828d217edde06466bda28de7ef877b","target":"graph","created_at":"2026-07-05T11:19:24Z","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/2502.05202/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Accelerating the inference of large language models (LLMs) is a critical challenge in generative AI. Speculative decoding (SD) methods offer substantial efficiency gains by generating multiple tokens using a single target forward pass. However, existing SD approaches require the drafter and target models to share the same vocabulary, thus limiting the pool of possible drafters, often necessitating the training of a drafter from scratch. We present three new SD methods that remove this shared-vocabulary constraint. All three methods preserve the target distribution (i.e., they are lossless) and","authors_text":"Daniel Korat, David Harel, Gaurav Jain, Jonathan Mamou, Moshe Berchansky, Moshe Wasserblat, Nadav Timor, Oren Pereg","cross_cats":["cs.AI","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-31T19:13:58Z","title":"Accelerating LLM Inference with Lossless Speculative Decoding Algorithms for Heterogeneous Vocabularies"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.05202","kind":"arxiv","version":3},"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:bc8d93496d4dcb0a30d7d299ce0fc018e75da3d39d694cf3c854874e70075a2e","target":"record","created_at":"2026-07-05T11:19:24Z","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":"8360f2ab6ab3ed597804e6789d7caff78cd6f2c2c11067b571f156cddede961c","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-31T19:13:58Z","title_canon_sha256":"2afeb64d2061d341b56798eae87af9cde5c7a6274cb492640e297b493fd9dbdc"},"schema_version":"1.0","source":{"id":"2502.05202","kind":"arxiv","version":3}},"canonical_sha256":"a5a3272c6cc6184c9cdb58a61b908b4b88aa0d18600c7f2f3f3654ee563db317","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a5a3272c6cc6184c9cdb58a61b908b4b88aa0d18600c7f2f3f3654ee563db317","first_computed_at":"2026-07-05T11:19:24.453899Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:19:24.453899Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"87pSf4CR0HCP86t24cJaTs8UcoJ2JXYxT1xQHxJMqS/r1Za+NkuPhjoKU53zCnXS8SydQk7CWd98bqdRFANQCg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:19:24.454403Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.05202","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bc8d93496d4dcb0a30d7d299ce0fc018e75da3d39d694cf3c854874e70075a2e","sha256:537c1cbf3b1c1a4033b442865e4f8fece3828d217edde06466bda28de7ef877b"],"state_sha256":"1d782aa220d491ebb2cf3157fc713f53a46d56f8de4bd54cc5d436d4075ce20b"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PAv1ZDYr15iu++CG2GyOWv3WZyvQ0/mAasA1+D4vyYWn/Sd/d7veX9uR9DvNQGICC05ZSkypuTZkdwH4Y/hXDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T02:45:03.904879Z","bundle_sha256":"f5780d3b95e562b10073db5f41071de8a7a66bd8abe8ab315326d223740d2bf2"}}