{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:LW535ZR3MRM2EGKNOAYHYO44M3","short_pith_number":"pith:LW535ZR3","canonical_record":{"source":{"id":"2404.00242","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-30T04:34:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c3ca4ae657582d75713600d38e28cf8746fddc15c4d7ea90905aadc238905faf","abstract_canon_sha256":"838624b09cf575d6146bc622aeae37dcaca9fdf3389cc3ba5ff376092f25e396"},"schema_version":"1.0"},"canonical_sha256":"5dbbbee63b6459a2194d70307c3b9c66d76b3f09205c299c9e9d890a3f8588b8","source":{"kind":"arxiv","id":"2404.00242","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.00242","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"arxiv_version","alias_value":"2404.00242v4","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.00242","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"pith_short_12","alias_value":"LW535ZR3MRM2","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"pith_short_16","alias_value":"LW535ZR3MRM2EGKN","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"pith_short_8","alias_value":"LW535ZR3","created_at":"2026-07-05T10:26:02Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:LW535ZR3MRM2EGKNOAYHYO44M3","target":"record","payload":{"canonical_record":{"source":{"id":"2404.00242","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-30T04:34:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c3ca4ae657582d75713600d38e28cf8746fddc15c4d7ea90905aadc238905faf","abstract_canon_sha256":"838624b09cf575d6146bc622aeae37dcaca9fdf3389cc3ba5ff376092f25e396"},"schema_version":"1.0"},"canonical_sha256":"5dbbbee63b6459a2194d70307c3b9c66d76b3f09205c299c9e9d890a3f8588b8","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:26:02.073696Z","signature_b64":"npXc/uKweFLQR4o3IohminAP7DM/8iU1kcip3v6s8FHOPf8K0dmRZWMN+FiN/3vhTaw/NCh7CWe+1SvRvh+6AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5dbbbee63b6459a2194d70307c3b9c66d76b3f09205c299c9e9d890a3f8588b8","last_reissued_at":"2026-07-05T10:26:02.072954Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:26:02.072954Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2404.00242","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:26:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Lj0slvQMDfB1WPl/aW2Dy/4Opcpzdih4e7pDG9X83fQ5TEZSr0bILmAra5+6aPx0NSgJieEpAa4wIhtJt8L/Dw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-20T06:06:54.555203Z"},"content_sha256":"be68ffc025b0a72172a9d8aae5deb10677534185abc15d7c62000fd89634871c","schema_version":"1.0","event_id":"sha256:be68ffc025b0a72172a9d8aae5deb10677534185abc15d7c62000fd89634871c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:LW535ZR3MRM2EGKNOAYHYO44M3","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"DeFT: Decoding with Flash Tree-attention for Efficient Tree-structured LLM Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Binhang Yuan, Jiaxuan You, Jinwei Yao, Kaiqi Chen, Kexun Zhang, Tao Lin, Zeke Wang","submitted_at":"2024-03-30T04:34:54Z","abstract_excerpt":"Large language models (LLMs) are increasingly employed for complex tasks that process multiple generation calls in a tree structure with shared prefixes of tokens, including few-shot prompting, multi-step reasoning, speculative decoding, etc. However, existing inference systems for tree-based applications are inefficient due to improper partitioning of queries and KV cache during attention calculation. This leads to two main issues: (1) a lack of memory access (IO) reuse for KV cache of shared prefixes, and (2) poor load balancing.As a result, there is redundant KV cache IO between GPU global "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.00242","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/2404.00242/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":"2026-07-05T10:26:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"i7ObP6zbg8SPFRBZz5LN7VD1D6H0sQLzJKsr0xkT/AX5iEocHL7odCuzesC+1KcpQWn2nqdAOPrPNIFDzEMmAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-20T06:06:54.556184Z"},"content_sha256":"96dde863fb072d8790aa8376a1e03395ace451d249bd4dc9b1b8263b654e799e","schema_version":"1.0","event_id":"sha256:96dde863fb072d8790aa8376a1e03395ace451d249bd4dc9b1b8263b654e799e"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/LW535ZR3MRM2EGKNOAYHYO44M3/bundle.json","state_url":"https://pith.science/pith/LW535ZR3MRM2EGKNOAYHYO44M3/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/LW535ZR3MRM2EGKNOAYHYO44M3/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":"2026-08-20T06:06:54Z","links":{"resolver":"https://pith.science/pith/LW535ZR3MRM2EGKNOAYHYO44M3","bundle":"https://pith.science/pith/LW535ZR3MRM2EGKNOAYHYO44M3/bundle.json","state":"https://pith.science/pith/LW535ZR3MRM2EGKNOAYHYO44M3/state.json","well_known_bundle":"https://pith.science/.well-known/pith/LW535ZR3MRM2EGKNOAYHYO44M3/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:LW535ZR3MRM2EGKNOAYHYO44M3","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":"838624b09cf575d6146bc622aeae37dcaca9fdf3389cc3ba5ff376092f25e396","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-30T04:34:54Z","title_canon_sha256":"c3ca4ae657582d75713600d38e28cf8746fddc15c4d7ea90905aadc238905faf"},"schema_version":"1.0","source":{"id":"2404.00242","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.00242","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"arxiv_version","alias_value":"2404.00242v4","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.00242","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"pith_short_12","alias_value":"LW535ZR3MRM2","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"pith_short_16","alias_value":"LW535ZR3MRM2EGKN","created_at":"2026-07-05T10:26:02Z"},{"alias_kind":"pith_short_8","alias_value":"LW535ZR3","created_at":"2026-07-05T10:26:02Z"}],"graph_snapshots":[{"event_id":"sha256:96dde863fb072d8790aa8376a1e03395ace451d249bd4dc9b1b8263b654e799e","target":"graph","created_at":"2026-07-05T10:26:02Z","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/2404.00242/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are increasingly employed for complex tasks that process multiple generation calls in a tree structure with shared prefixes of tokens, including few-shot prompting, multi-step reasoning, speculative decoding, etc. However, existing inference systems for tree-based applications are inefficient due to improper partitioning of queries and KV cache during attention calculation. This leads to two main issues: (1) a lack of memory access (IO) reuse for KV cache of shared prefixes, and (2) poor load balancing.As a result, there is redundant KV cache IO between GPU global ","authors_text":"Binhang Yuan, Jiaxuan You, Jinwei Yao, Kaiqi Chen, Kexun Zhang, Tao Lin, Zeke Wang","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-30T04:34:54Z","title":"DeFT: Decoding with Flash Tree-attention for Efficient Tree-structured LLM Inference"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.00242","kind":"arxiv","version":4},"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:be68ffc025b0a72172a9d8aae5deb10677534185abc15d7c62000fd89634871c","target":"record","created_at":"2026-07-05T10:26:02Z","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":"838624b09cf575d6146bc622aeae37dcaca9fdf3389cc3ba5ff376092f25e396","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-30T04:34:54Z","title_canon_sha256":"c3ca4ae657582d75713600d38e28cf8746fddc15c4d7ea90905aadc238905faf"},"schema_version":"1.0","source":{"id":"2404.00242","kind":"arxiv","version":4}},"canonical_sha256":"5dbbbee63b6459a2194d70307c3b9c66d76b3f09205c299c9e9d890a3f8588b8","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"5dbbbee63b6459a2194d70307c3b9c66d76b3f09205c299c9e9d890a3f8588b8","first_computed_at":"2026-07-05T10:26:02.072954Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:26:02.072954Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"npXc/uKweFLQR4o3IohminAP7DM/8iU1kcip3v6s8FHOPf8K0dmRZWMN+FiN/3vhTaw/NCh7CWe+1SvRvh+6AQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:26:02.073696Z","signed_message":"canonical_sha256_bytes"},"source_id":"2404.00242","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:be68ffc025b0a72172a9d8aae5deb10677534185abc15d7c62000fd89634871c","sha256:96dde863fb072d8790aa8376a1e03395ace451d249bd4dc9b1b8263b654e799e"],"state_sha256":"fd75a3804d569b82d3c91fc02241440f0c8d8e43f4985e549c295854137f5d8a"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FIvLvKN+yszvrmaBNeNGoRo3Yd/CS6ZTU/655l81wJwnatVEH2NLq942w8yfRSn13cVcKcIUWUmbPA+QZuN8BQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-20T06:06:54.561903Z","bundle_sha256":"ebdd05084fdd5c953fbdd05d5788b8bc39ec213afc59df51cea3a1529bc0050e"}}