{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:PCUFW5E7XVSLA2TRPZ5VJ3TLRU","short_pith_number":"pith:PCUFW5E7","schema_version":"1.0","canonical_sha256":"78a85b749fbd64b06a717e7b54ee6b8d03703f19417af4a1c248c5b5d1d619d1","source":{"kind":"arxiv","id":"2607.04763","version":1},"attestation_state":"computed","paper":{"title":"Multi-Turn On-Policy Distillation with Prefix Replay","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Baohao Liao, Christof Monz, Furu Wei, HanZe Dong, Li Dong, Xinxing Xu","submitted_at":"2026-07-06T07:56:53Z","abstract_excerpt":"We study on-policy distillation (OPD) for agentic tasks, where an LLM agent interacts with an environment over multiple turns and a student imitates a teacher over these multi-turn interaction histories. Fully online OPD is costly because each update requires fresh student rollouts through the environment and teacher queries at visited histories. We propose Replayed-Prefix On-Policy Distillation (ReOPD), an off-environment alternative that reuses pre-collected teacher trajectories as replayed prefixes: the student acts at selected steps, while the teacher provides dense per-step supervision wi"},"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":"2607.04763","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-07-06T07:56:53Z","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"title_canon_sha256":"666eccb0e46b79d3f9b4033857629bada6174b3ddb9ca1648a8cb626347df36c","abstract_canon_sha256":"ab9128020cc5be59bee7198f4758ca051a41d8a7899c969a5a0c7804af3339ff"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:20:02.207600Z","signature_b64":"0lC5VpLYsniaiO8V2Mg4tAcItPSfDU8Q4CW1Ur0iygqBlJ53JQ8bvgxzn0+yTj8oBiOxq2alllCQLB9ZJ9rNAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"78a85b749fbd64b06a717e7b54ee6b8d03703f19417af4a1c248c5b5d1d619d1","last_reissued_at":"2026-07-07T02:20:02.206891Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:20:02.206891Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multi-Turn On-Policy Distillation with Prefix Replay","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Baohao Liao, Christof Monz, Furu Wei, HanZe Dong, Li Dong, Xinxing Xu","submitted_at":"2026-07-06T07:56:53Z","abstract_excerpt":"We study on-policy distillation (OPD) for agentic tasks, where an LLM agent interacts with an environment over multiple turns and a student imitates a teacher over these multi-turn interaction histories. Fully online OPD is costly because each update requires fresh student rollouts through the environment and teacher queries at visited histories. We propose Replayed-Prefix On-Policy Distillation (ReOPD), an off-environment alternative that reuses pre-collected teacher trajectories as replayed prefixes: the student acts at selected steps, while the teacher provides dense per-step supervision wi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.04763","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/2607.04763/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":"2607.04763","created_at":"2026-07-07T02:20:02.207004+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.04763v1","created_at":"2026-07-07T02:20:02.207004+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.04763","created_at":"2026-07-07T02:20:02.207004+00:00"},{"alias_kind":"pith_short_12","alias_value":"PCUFW5E7XVSL","created_at":"2026-07-07T02:20:02.207004+00:00"},{"alias_kind":"pith_short_16","alias_value":"PCUFW5E7XVSLA2TR","created_at":"2026-07-07T02:20:02.207004+00:00"},{"alias_kind":"pith_short_8","alias_value":"PCUFW5E7","created_at":"2026-07-07T02:20:02.207004+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/PCUFW5E7XVSLA2TRPZ5VJ3TLRU","json":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU.json","graph_json":"https://pith.science/api/pith-number/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/graph.json","events_json":"https://pith.science/api/pith-number/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/events.json","paper":"https://pith.science/paper/PCUFW5E7"},"agent_actions":{"view_html":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU","download_json":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU.json","view_paper":"https://pith.science/paper/PCUFW5E7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.04763&json=true","fetch_graph":"https://pith.science/api/pith-number/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/graph.json","fetch_events":"https://pith.science/api/pith-number/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/action/storage_attestation","attest_author":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/action/author_attestation","sign_citation":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/action/citation_signature","submit_replication":"https://pith.science/pith/PCUFW5E7XVSLA2TRPZ5VJ3TLRU/action/replication_record"}},"created_at":"2026-07-07T02:20:02.207004+00:00","updated_at":"2026-07-07T02:20:02.207004+00:00"}