{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YW2IHBRHK4DJUYJDBGSV4EJR47","short_pith_number":"pith:YW2IHBRH","schema_version":"1.0","canonical_sha256":"c5b483862757069a612309a55e1131e7ef215b817b9f2df3226acd270f9359f4","source":{"kind":"arxiv","id":"2606.09165","version":1},"attestation_state":"computed","paper":{"title":"Reliable to Expressive: A Curriculum for Rubric-Following Safety Judges","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Hyeji Choi, Minwoo Kim, YongTaek Lim","submitted_at":"2026-06-08T08:02:57Z","abstract_excerpt":"Safety judges are increasingly deployed to evaluate model outputs against evolving criteria, yet recent meta-evaluation work shows they remain brittle under prompt and rubric variation, with false negative-rate swings of up to 0.24 reported for stylistic perturbations alone. We argue that safety judgment is fundamentally a rubric-following problem: a robust judge must apply the given evaluation criteria consistently across rubric formulations rather than memorize one specific template. We propose a training strategy that combines (i) instance-conditioned dynamic rubrics generated from prompt-r"},"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":"2606.09165","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-08T08:02:57Z","cross_cats_sorted":[],"title_canon_sha256":"7c58a4692ca39f7df48fe000b337f4435df62fb0d0a52cdebe4e96514e747588","abstract_canon_sha256":"197e113efb1a79125b91736bc33894d9e94f495493242e5b14a1d0e9998504d2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:08:04.053133Z","signature_b64":"ISuJbGv2VBffIS+Z2vetG7CwXO9PW3Hi/PYssyyZi9zZoEC5Jfkb8uTVzYuNubVjI5kXLpQOyG8fnn0IOoDpDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5b483862757069a612309a55e1131e7ef215b817b9f2df3226acd270f9359f4","last_reissued_at":"2026-06-09T02:08:04.052264Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:08:04.052264Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reliable to Expressive: A Curriculum for Rubric-Following Safety Judges","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Hyeji Choi, Minwoo Kim, YongTaek Lim","submitted_at":"2026-06-08T08:02:57Z","abstract_excerpt":"Safety judges are increasingly deployed to evaluate model outputs against evolving criteria, yet recent meta-evaluation work shows they remain brittle under prompt and rubric variation, with false negative-rate swings of up to 0.24 reported for stylistic perturbations alone. We argue that safety judgment is fundamentally a rubric-following problem: a robust judge must apply the given evaluation criteria consistently across rubric formulations rather than memorize one specific template. We propose a training strategy that combines (i) instance-conditioned dynamic rubrics generated from prompt-r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09165","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/2606.09165/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":"2606.09165","created_at":"2026-06-09T02:08:04.052420+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.09165v1","created_at":"2026-06-09T02:08:04.052420+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09165","created_at":"2026-06-09T02:08:04.052420+00:00"},{"alias_kind":"pith_short_12","alias_value":"YW2IHBRHK4DJ","created_at":"2026-06-09T02:08:04.052420+00:00"},{"alias_kind":"pith_short_16","alias_value":"YW2IHBRHK4DJUYJD","created_at":"2026-06-09T02:08:04.052420+00:00"},{"alias_kind":"pith_short_8","alias_value":"YW2IHBRH","created_at":"2026-06-09T02:08:04.052420+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/YW2IHBRHK4DJUYJDBGSV4EJR47","json":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47.json","graph_json":"https://pith.science/api/pith-number/YW2IHBRHK4DJUYJDBGSV4EJR47/graph.json","events_json":"https://pith.science/api/pith-number/YW2IHBRHK4DJUYJDBGSV4EJR47/events.json","paper":"https://pith.science/paper/YW2IHBRH"},"agent_actions":{"view_html":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47","download_json":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47.json","view_paper":"https://pith.science/paper/YW2IHBRH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.09165&json=true","fetch_graph":"https://pith.science/api/pith-number/YW2IHBRHK4DJUYJDBGSV4EJR47/graph.json","fetch_events":"https://pith.science/api/pith-number/YW2IHBRHK4DJUYJDBGSV4EJR47/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47/action/storage_attestation","attest_author":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47/action/author_attestation","sign_citation":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47/action/citation_signature","submit_replication":"https://pith.science/pith/YW2IHBRHK4DJUYJDBGSV4EJR47/action/replication_record"}},"created_at":"2026-06-09T02:08:04.052420+00:00","updated_at":"2026-06-09T02:08:04.052420+00:00"}