{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7VXK2XQJ5FZCQPHQJOY2WKSRRS","short_pith_number":"pith:7VXK2XQJ","schema_version":"1.0","canonical_sha256":"fd6ead5e09e972283cf04bb1ab2a518cb20fc2a22e7bc45d5bea7a1dd146459f","source":{"kind":"arxiv","id":"2608.10209","version":1},"attestation_state":"computed","paper":{"title":"Evaluation-Conditioned Training: Teaching Models to Generalize to Stronger Oversight Regimes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Alec Harris, Archie Chaudhury, Kasey Corra, Yixiong Hao","submitted_at":"2026-08-10T20:22:13Z","abstract_excerpt":"Feedback signals used to train Large Language Models (LLMs) are the primary driver of their behavior and our main lever for instilling alignment with human values and objectives. However, a key limitation of current post-training methods is the inability of human annotators and automated reward functions to faithfully capture the feedback we would like to give. We introduce Evaluation-Conditioned Training (ECT), a post-training framework that uses natural language to condition each training sample on the fidelity of the feedback we provide and then elicits the desired behavior by conditioning "},"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":"2608.10209","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-10T20:22:13Z","cross_cats_sorted":[],"title_canon_sha256":"3b1e8a38b6e85170cf77c43bdc1a719ec5818ea7839968a67f7fbac35e3cb7dd","abstract_canon_sha256":"2f114141e82afbc22c0e4a32f932ed7e4f09a94f287b4cb2812c39c1a20c5877"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-12T00:23:04.332253Z","signature_b64":"qq44uUZYued/QGCJ2QlQ4Ysi1QcUWqZ1fxiJ4N0/r/aSGhCHODnBfe/axic5Sj8QWlcYnL7L7wV+rRH7YdyIAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fd6ead5e09e972283cf04bb1ab2a518cb20fc2a22e7bc45d5bea7a1dd146459f","last_reissued_at":"2026-08-12T00:23:04.330802Z","signature_status":"signed_v1","first_computed_at":"2026-08-12T00:23:04.330802Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Evaluation-Conditioned Training: Teaching Models to Generalize to Stronger Oversight Regimes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Alec Harris, Archie Chaudhury, Kasey Corra, Yixiong Hao","submitted_at":"2026-08-10T20:22:13Z","abstract_excerpt":"Feedback signals used to train Large Language Models (LLMs) are the primary driver of their behavior and our main lever for instilling alignment with human values and objectives. However, a key limitation of current post-training methods is the inability of human annotators and automated reward functions to faithfully capture the feedback we would like to give. We introduce Evaluation-Conditioned Training (ECT), a post-training framework that uses natural language to condition each training sample on the fidelity of the feedback we provide and then elicits the desired behavior by conditioning "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.10209","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/2608.10209/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":"2608.10209","created_at":"2026-08-12T00:23:04.333429+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.10209v1","created_at":"2026-08-12T00:23:04.333429+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.10209","created_at":"2026-08-12T00:23:04.333429+00:00"},{"alias_kind":"pith_short_12","alias_value":"7VXK2XQJ5FZC","created_at":"2026-08-12T00:23:04.333429+00:00"},{"alias_kind":"pith_short_16","alias_value":"7VXK2XQJ5FZCQPHQ","created_at":"2026-08-12T00:23:04.333429+00:00"},{"alias_kind":"pith_short_8","alias_value":"7VXK2XQJ","created_at":"2026-08-12T00:23:04.333429+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/7VXK2XQJ5FZCQPHQJOY2WKSRRS","json":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS.json","graph_json":"https://pith.science/api/pith-number/7VXK2XQJ5FZCQPHQJOY2WKSRRS/graph.json","events_json":"https://pith.science/api/pith-number/7VXK2XQJ5FZCQPHQJOY2WKSRRS/events.json","paper":"https://pith.science/paper/7VXK2XQJ"},"agent_actions":{"view_html":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS","download_json":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS.json","view_paper":"https://pith.science/paper/7VXK2XQJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.10209&json=true","fetch_graph":"https://pith.science/api/pith-number/7VXK2XQJ5FZCQPHQJOY2WKSRRS/graph.json","fetch_events":"https://pith.science/api/pith-number/7VXK2XQJ5FZCQPHQJOY2WKSRRS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS/action/storage_attestation","attest_author":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS/action/author_attestation","sign_citation":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS/action/citation_signature","submit_replication":"https://pith.science/pith/7VXK2XQJ5FZCQPHQJOY2WKSRRS/action/replication_record"}},"created_at":"2026-08-12T00:23:04.333429+00:00","updated_at":"2026-08-12T00:23:04.333429+00:00"}