{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZHQLJ4XORNFAEKGYJMD5AHT7IU","short_pith_number":"pith:ZHQLJ4XO","schema_version":"1.0","canonical_sha256":"c9e0b4f2ee8b4a0228d84b07d01e7f453a1ed066dd75cb1268896f5bdf405ba8","source":{"kind":"arxiv","id":"2606.18967","version":1},"attestation_state":"computed","paper":{"title":"EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Amir Gholami, Coleman Hooper, Donghoon Kim, Harman Singh, Hyung Il Koo, Kevin Galim, Minjae Lee, Minseo Kim, Seunghyuk Oh, Wonjun Kang","submitted_at":"2026-06-17T11:51:06Z","abstract_excerpt":"Reinforcement learning (RL) has become a representative post-training paradigm for LLMs, enabling strong reasoning and agentic capabilities. However, rollout generation remains a dominant latency bottleneck because autoregressive sampling decodes responses sequentially and a small number of long-tailed generations often determine completion time. Speculative decoding (SD) offers a natural way to address this bottleneck, as it is a well-established technique for serving fixed LLMs that reduces latency by rapidly drafting tokens and accepting them through parallel verification while preserving t"},"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":"2606.18967","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-17T11:51:06Z","cross_cats_sorted":[],"title_canon_sha256":"9d7582a2880b2deaaac8e77ffdf96afb6fd97df2e299fc86350abd5af8133d48","abstract_canon_sha256":"c281528103e7dd0912ec946fc556675666507dcdb01fd260cd2682ac40d78151"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:11:53.346602Z","signature_b64":"5Eslh9qi5s+Bq3fhbOdWQ7LZ+SrmXykKDIZAcu5iNTC2M6nh9hhrhvFp7kTWnr8OyoVux7X+cKyY1CSSTeQQBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c9e0b4f2ee8b4a0228d84b07d01e7f453a1ed066dd75cb1268896f5bdf405ba8","last_reissued_at":"2026-06-19T16:11:53.346112Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:11:53.346112Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Amir Gholami, Coleman Hooper, Donghoon Kim, Harman Singh, Hyung Il Koo, Kevin Galim, Minjae Lee, Minseo Kim, Seunghyuk Oh, Wonjun Kang","submitted_at":"2026-06-17T11:51:06Z","abstract_excerpt":"Reinforcement learning (RL) has become a representative post-training paradigm for LLMs, enabling strong reasoning and agentic capabilities. However, rollout generation remains a dominant latency bottleneck because autoregressive sampling decodes responses sequentially and a small number of long-tailed generations often determine completion time. Speculative decoding (SD) offers a natural way to address this bottleneck, as it is a well-established technique for serving fixed LLMs that reduces latency by rapidly drafting tokens and accepting them through parallel verification while preserving t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.18967","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/2606.18967/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":"2606.18967","created_at":"2026-06-19T16:11:53.346175+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.18967v1","created_at":"2026-06-19T16:11:53.346175+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.18967","created_at":"2026-06-19T16:11:53.346175+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZHQLJ4XORNFA","created_at":"2026-06-19T16:11:53.346175+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZHQLJ4XORNFAEKGY","created_at":"2026-06-19T16:11:53.346175+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZHQLJ4XO","created_at":"2026-06-19T16:11:53.346175+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/ZHQLJ4XORNFAEKGYJMD5AHT7IU","json":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU.json","graph_json":"https://pith.science/api/pith-number/ZHQLJ4XORNFAEKGYJMD5AHT7IU/graph.json","events_json":"https://pith.science/api/pith-number/ZHQLJ4XORNFAEKGYJMD5AHT7IU/events.json","paper":"https://pith.science/paper/ZHQLJ4XO"},"agent_actions":{"view_html":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU","download_json":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU.json","view_paper":"https://pith.science/paper/ZHQLJ4XO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.18967&json=true","fetch_graph":"https://pith.science/api/pith-number/ZHQLJ4XORNFAEKGYJMD5AHT7IU/graph.json","fetch_events":"https://pith.science/api/pith-number/ZHQLJ4XORNFAEKGYJMD5AHT7IU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU/action/storage_attestation","attest_author":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU/action/author_attestation","sign_citation":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU/action/citation_signature","submit_replication":"https://pith.science/pith/ZHQLJ4XORNFAEKGYJMD5AHT7IU/action/replication_record"}},"created_at":"2026-06-19T16:11:53.346175+00:00","updated_at":"2026-06-19T16:11:53.346175+00:00"}