{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UBC674LNXM3WO3RFPQYDAQ6PXA","short_pith_number":"pith:UBC674LN","schema_version":"1.0","canonical_sha256":"a045eff16dbb37676e257c303043cfb80b0c89d9096df5d02da49625a7eadaa2","source":{"kind":"arxiv","id":"2608.01651","version":1},"attestation_state":"computed","paper":{"title":"Bole: Efficient Tree Speculation for Hybrid-Attention Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.DC","authors_text":"Chengying Huan, Chen Tian, Chiran You, Fangxin Liu, Jiajun Zheng, Jie Zhang, Juelu Zhang, Li Wang, Xiabao Wu, Yi Su, Yongchao Liu, Zhan Qiu","submitted_at":"2026-08-03T03:43:14Z","abstract_excerpt":"Hybrid-attention large language models combine full attention with recurrent linear attention to reduce long-context inference costs, yet their autoregressive decoding remains memory-bound. Tree speculative decoding offers an attractive acceleration path, but existing tree-speculation systems are designed around the key--value caches of full-attention models. On hybrid models, they traverse recurrent layers branch by branch and materialize a full state for every proposal node, causing verification latency and transient memory to scale poorly with tree and batch sizes. We present Bole, a kernel"},"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":"2608.01651","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2026-08-03T03:43:14Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"ce46f9a5bc9059583d4150660bec1b2170a4d3bc21138e68310dac660dc39ce9","abstract_canon_sha256":"f9490a5f5052389c2756ffdf0f62fb5e6fb19ace01eacac48670cd6098683ee2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-04T02:06:20.382905Z","signature_b64":"MHawhyauUT+gdjPAVX5T//EwvAmJwGjKiPgVTXK8TtmaXyys4GylgFrTFlQBfzJOspcgH0ggQWl7X84RVC5jAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a045eff16dbb37676e257c303043cfb80b0c89d9096df5d02da49625a7eadaa2","last_reissued_at":"2026-08-04T02:06:20.381374Z","signature_status":"signed_v1","first_computed_at":"2026-08-04T02:06:20.381374Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bole: Efficient Tree Speculation for Hybrid-Attention Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.DC","authors_text":"Chengying Huan, Chen Tian, Chiran You, Fangxin Liu, Jiajun Zheng, Jie Zhang, Juelu Zhang, Li Wang, Xiabao Wu, Yi Su, Yongchao Liu, Zhan Qiu","submitted_at":"2026-08-03T03:43:14Z","abstract_excerpt":"Hybrid-attention large language models combine full attention with recurrent linear attention to reduce long-context inference costs, yet their autoregressive decoding remains memory-bound. Tree speculative decoding offers an attractive acceleration path, but existing tree-speculation systems are designed around the key--value caches of full-attention models. On hybrid models, they traverse recurrent layers branch by branch and materialize a full state for every proposal node, causing verification latency and transient memory to scale poorly with tree and batch sizes. We present Bole, a kernel"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.01651","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/2608.01651/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":"2608.01651","created_at":"2026-08-04T02:06:20.382897+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.01651v1","created_at":"2026-08-04T02:06:20.382897+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.01651","created_at":"2026-08-04T02:06:20.382897+00:00"},{"alias_kind":"pith_short_12","alias_value":"UBC674LNXM3W","created_at":"2026-08-04T02:06:20.382897+00:00"},{"alias_kind":"pith_short_16","alias_value":"UBC674LNXM3WO3RF","created_at":"2026-08-04T02:06:20.382897+00:00"},{"alias_kind":"pith_short_8","alias_value":"UBC674LN","created_at":"2026-08-04T02:06:20.382897+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/UBC674LNXM3WO3RFPQYDAQ6PXA","json":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA.json","graph_json":"https://pith.science/api/pith-number/UBC674LNXM3WO3RFPQYDAQ6PXA/graph.json","events_json":"https://pith.science/api/pith-number/UBC674LNXM3WO3RFPQYDAQ6PXA/events.json","paper":"https://pith.science/paper/UBC674LN"},"agent_actions":{"view_html":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA","download_json":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA.json","view_paper":"https://pith.science/paper/UBC674LN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.01651&json=true","fetch_graph":"https://pith.science/api/pith-number/UBC674LNXM3WO3RFPQYDAQ6PXA/graph.json","fetch_events":"https://pith.science/api/pith-number/UBC674LNXM3WO3RFPQYDAQ6PXA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA/action/storage_attestation","attest_author":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA/action/author_attestation","sign_citation":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA/action/citation_signature","submit_replication":"https://pith.science/pith/UBC674LNXM3WO3RFPQYDAQ6PXA/action/replication_record"}},"created_at":"2026-08-04T02:06:20.382897+00:00","updated_at":"2026-08-04T02:06:20.382897+00:00"}