{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZD56OO7JFXC3PRZ452D2HRASRD","short_pith_number":"pith:ZD56OO7J","schema_version":"1.0","canonical_sha256":"c8fbe73be92dc5b7c73cee87a3c41288d3e318ad1a58f5adfb7bb7ef3a40bf78","source":{"kind":"arxiv","id":"2607.17524","version":1},"attestation_state":"computed","paper":{"title":"Token-Level Off-Policy Learning for Faithful Generation Under Distribution Shift","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Deqing Fu, Gustavo Lucas Carvalho, Robin Jia, Zitong Huang","submitted_at":"2026-07-20T03:57:22Z","abstract_excerpt":"We propose Token-Level Off-Policy Labeling (TOPL), an off-policy training paradigm that reframes post-training as a token-level correctness prediction task. Our key intuition is that by training the model to distinguish good and bad tokens in a response, we naturally guide the model towards generating good tokens, while avoiding the pitfalls that come with directly training the model to generate off-policy tokens. Experiments on document summarization tasks show that TOPL achieves strong out-of-distribution generalization across 11 datasets against a diverse set of sequence-level and token-lev"},"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":"2607.17524","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-20T03:57:22Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"0ba5bf7123c7a5f13182c4c58ce60065149f6ee70f7ce8705173106426b4c498","abstract_canon_sha256":"2607ef46c8932f8baa9dc124dd9d75d893f2056e98c3ec0d6268310463f7381b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-21T01:21:50.463782Z","signature_b64":"u/74CPchmpuWtltyrLYzcvRTZOz16PE+0Zl97Rh7LVKCld3F1tU7SNa2TF5786QomJ+XV2WyxTT65fk0mUWJCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c8fbe73be92dc5b7c73cee87a3c41288d3e318ad1a58f5adfb7bb7ef3a40bf78","last_reissued_at":"2026-07-21T01:21:50.462924Z","signature_status":"signed_v1","first_computed_at":"2026-07-21T01:21:50.462924Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Token-Level Off-Policy Learning for Faithful Generation Under Distribution Shift","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Deqing Fu, Gustavo Lucas Carvalho, Robin Jia, Zitong Huang","submitted_at":"2026-07-20T03:57:22Z","abstract_excerpt":"We propose Token-Level Off-Policy Labeling (TOPL), an off-policy training paradigm that reframes post-training as a token-level correctness prediction task. Our key intuition is that by training the model to distinguish good and bad tokens in a response, we naturally guide the model towards generating good tokens, while avoiding the pitfalls that come with directly training the model to generate off-policy tokens. Experiments on document summarization tasks show that TOPL achieves strong out-of-distribution generalization across 11 datasets against a diverse set of sequence-level and token-lev"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.17524","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/2607.17524/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":"2607.17524","created_at":"2026-07-21T01:21:50.463372+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.17524v1","created_at":"2026-07-21T01:21:50.463372+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.17524","created_at":"2026-07-21T01:21:50.463372+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZD56OO7JFXC3","created_at":"2026-07-21T01:21:50.463372+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZD56OO7JFXC3PRZ4","created_at":"2026-07-21T01:21:50.463372+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZD56OO7J","created_at":"2026-07-21T01:21:50.463372+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/ZD56OO7JFXC3PRZ452D2HRASRD","json":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD.json","graph_json":"https://pith.science/api/pith-number/ZD56OO7JFXC3PRZ452D2HRASRD/graph.json","events_json":"https://pith.science/api/pith-number/ZD56OO7JFXC3PRZ452D2HRASRD/events.json","paper":"https://pith.science/paper/ZD56OO7J"},"agent_actions":{"view_html":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD","download_json":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD.json","view_paper":"https://pith.science/paper/ZD56OO7J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.17524&json=true","fetch_graph":"https://pith.science/api/pith-number/ZD56OO7JFXC3PRZ452D2HRASRD/graph.json","fetch_events":"https://pith.science/api/pith-number/ZD56OO7JFXC3PRZ452D2HRASRD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD/action/storage_attestation","attest_author":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD/action/author_attestation","sign_citation":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD/action/citation_signature","submit_replication":"https://pith.science/pith/ZD56OO7JFXC3PRZ452D2HRASRD/action/replication_record"}},"created_at":"2026-07-21T01:21:50.463372+00:00","updated_at":"2026-07-21T01:21:50.463372+00:00"}