{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:R3KVYEFYTS4G3F7FCUZUZYVFSQ","short_pith_number":"pith:R3KVYEFY","schema_version":"1.0","canonical_sha256":"8ed55c10b89cb86d97e515334ce2a5943cf48da0cc33761a0e756b918fcab1f5","source":{"kind":"arxiv","id":"2607.04574","version":1},"attestation_state":"computed","paper":{"title":"A Few Teacher Steps Go a Long Way: Cost-Efficient On-Policy Data Augmentation for Agent Post-Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jiayi Cheng, Jose Blanchet, Junze Ye, Miao Lu, Michal Mankowski, Mohsen Bayati","submitted_at":"2026-07-06T01:02:30Z","abstract_excerpt":"For LLM agents, supervised fine-tuning is not only about teacher labels' quality, but also about which interaction contexts those labels condition on. Pure behavioral cloning uses full teacher demonstrations, creating a mismatch between teacher-induced contexts seen in training and student-induced contexts encountered at test time. Recent work addresses this mismatch by querying a teacher at contexts reached by the student, often with increasingly elaborate filtering of the teacher's continuations. We instead frame on-policy data construction as a budget-allocation problem: under matched super"},"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":"2607.04574","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-06T01:02:30Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"51145b420cd400d5da64c1f976a1fc20d67e7549ffd5b5ffdd8951de68af83b9","abstract_canon_sha256":"04f3db6172d0e548aa491f65d78e3b1a74cfe1904a28f59ee8a1f024d6abe27f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:19:25.512428Z","signature_b64":"S48CGYtWy+MW9kbj5ulC4LZ6TRQSqzxsrqr/8bd2Mjhc6pJPe+CfjtOfdzzT+hbWyJYSOHVvce3dDmhf40l4DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ed55c10b89cb86d97e515334ce2a5943cf48da0cc33761a0e756b918fcab1f5","last_reissued_at":"2026-07-07T02:19:25.511466Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:19:25.511466Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Few Teacher Steps Go a Long Way: Cost-Efficient On-Policy Data Augmentation for Agent Post-Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jiayi Cheng, Jose Blanchet, Junze Ye, Miao Lu, Michal Mankowski, Mohsen Bayati","submitted_at":"2026-07-06T01:02:30Z","abstract_excerpt":"For LLM agents, supervised fine-tuning is not only about teacher labels' quality, but also about which interaction contexts those labels condition on. Pure behavioral cloning uses full teacher demonstrations, creating a mismatch between teacher-induced contexts seen in training and student-induced contexts encountered at test time. Recent work addresses this mismatch by querying a teacher at contexts reached by the student, often with increasingly elaborate filtering of the teacher's continuations. We instead frame on-policy data construction as a budget-allocation problem: under matched super"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.04574","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/2607.04574/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":"2607.04574","created_at":"2026-07-07T02:19:25.511635+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.04574v1","created_at":"2026-07-07T02:19:25.511635+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.04574","created_at":"2026-07-07T02:19:25.511635+00:00"},{"alias_kind":"pith_short_12","alias_value":"R3KVYEFYTS4G","created_at":"2026-07-07T02:19:25.511635+00:00"},{"alias_kind":"pith_short_16","alias_value":"R3KVYEFYTS4G3F7F","created_at":"2026-07-07T02:19:25.511635+00:00"},{"alias_kind":"pith_short_8","alias_value":"R3KVYEFY","created_at":"2026-07-07T02:19:25.511635+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/R3KVYEFYTS4G3F7FCUZUZYVFSQ","json":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ.json","graph_json":"https://pith.science/api/pith-number/R3KVYEFYTS4G3F7FCUZUZYVFSQ/graph.json","events_json":"https://pith.science/api/pith-number/R3KVYEFYTS4G3F7FCUZUZYVFSQ/events.json","paper":"https://pith.science/paper/R3KVYEFY"},"agent_actions":{"view_html":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ","download_json":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ.json","view_paper":"https://pith.science/paper/R3KVYEFY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.04574&json=true","fetch_graph":"https://pith.science/api/pith-number/R3KVYEFYTS4G3F7FCUZUZYVFSQ/graph.json","fetch_events":"https://pith.science/api/pith-number/R3KVYEFYTS4G3F7FCUZUZYVFSQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ/action/storage_attestation","attest_author":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ/action/author_attestation","sign_citation":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ/action/citation_signature","submit_replication":"https://pith.science/pith/R3KVYEFYTS4G3F7FCUZUZYVFSQ/action/replication_record"}},"created_at":"2026-07-07T02:19:25.511635+00:00","updated_at":"2026-07-07T02:19:25.511635+00:00"}