{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:R6APKS7T34X7NKFOYYN5A6CNVD","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"4132ef63275c09cf4a368e95f371a762ba9a20ee2aec859a848c05f2386ddee6","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-20T07:32:22Z","title_canon_sha256":"016d3b075de16b680a5ccb96eac499c6725bb6a10992f4fc17c241d1eb0fe789"},"schema_version":"1.0","source":{"id":"2602.18037","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.18037","created_at":"2026-07-07T00:15:52Z"},{"alias_kind":"arxiv_version","alias_value":"2602.18037v2","created_at":"2026-07-07T00:15:52Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.18037","created_at":"2026-07-07T00:15:52Z"},{"alias_kind":"pith_short_12","alias_value":"R6APKS7T34X7","created_at":"2026-07-07T00:15:52Z"},{"alias_kind":"pith_short_16","alias_value":"R6APKS7T34X7NKFO","created_at":"2026-07-07T00:15:52Z"},{"alias_kind":"pith_short_8","alias_value":"R6APKS7T","created_at":"2026-07-07T00:15:52Z"}],"graph_snapshots":[{"event_id":"sha256:783a7903f1dbbc873bb81502666cd0f0542afc1dd82a169f589841eed0dbb16b","target":"graph","created_at":"2026-07-07T00:15:52Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2602.18037/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement Learning from Human Feedback (RLHF) or Verifiable Rewards (RLVR) are two key steps in the post-training of modern Language Models (LMs). A common problem is reward hacking, where the policy may exploit inaccuracies of the reward and learn an unintended behavior. Most previous works address this by limiting the policy update with a Kullback-Leibler (KL) penalty towards a reference model. We propose a different framing: Train the LM in a way that biases policy updates towards regions in which the reward is more accurate. First, we derive a theoretical connection between the accurac","authors_text":"Johannes Ackermann, Masashi Sugiyama, Michael Noukhovitch, Takashi Ishida","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-20T07:32:22Z","title":"Gradient Regularization Mitigates Reward Hacking in Reinforcement Learning from Human Feedback and Verifiable Rewards"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.18037","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ec31afb78c999d4d16fd7f0e293d8d01af62e02a9443db4be37afa2570bf6eec","target":"record","created_at":"2026-07-07T00:15:52Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"4132ef63275c09cf4a368e95f371a762ba9a20ee2aec859a848c05f2386ddee6","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-20T07:32:22Z","title_canon_sha256":"016d3b075de16b680a5ccb96eac499c6725bb6a10992f4fc17c241d1eb0fe789"},"schema_version":"1.0","source":{"id":"2602.18037","kind":"arxiv","version":2}},"canonical_sha256":"8f80f54bf3df2ff6a8aec61bd0784da8ef0c8a0f060c2524ed25fcc582aa2e80","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8f80f54bf3df2ff6a8aec61bd0784da8ef0c8a0f060c2524ed25fcc582aa2e80","first_computed_at":"2026-07-07T00:15:52.603114Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-07T00:15:52.603114Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"bWjQan/2HS30Uq+6pwRutjnlaMNyV+x2gPZOP17P7BJv0NROx5fgVqALtHhMS1qBUyYTuiPL6ulkmwHd2FqnDQ==","signature_status":"signed_v1","signed_at":"2026-07-07T00:15:52.603871Z","signed_message":"canonical_sha256_bytes"},"source_id":"2602.18037","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ec31afb78c999d4d16fd7f0e293d8d01af62e02a9443db4be37afa2570bf6eec","sha256:783a7903f1dbbc873bb81502666cd0f0542afc1dd82a169f589841eed0dbb16b"],"state_sha256":"1fbc6fc101880b2da73a77f1b5a69e39e79fb779e68c1d553c668cf2125edaf0"}