{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:YOVLSOXRJP3ZCP4CM7EDR2IQYL","short_pith_number":"pith:YOVLSOXR","canonical_record":{"source":{"id":"2606.28955","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-27T14:53:03Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d39defa5e227f6d7d8b05c8d32215a837b1be0593b2f5f895d3f49d616ebd3fd","abstract_canon_sha256":"7712cb537c9dec24ad00735ac2044a5827105c31f6270c727fff42f273bc403e"},"schema_version":"1.0"},"canonical_sha256":"c3aab93af14bf7913f8267c838e910c2d024b99dbc4926f15a78d5f55146d01c","source":{"kind":"arxiv","id":"2606.28955","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.28955","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"arxiv_version","alias_value":"2606.28955v1","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.28955","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"pith_short_12","alias_value":"YOVLSOXRJP3Z","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"pith_short_16","alias_value":"YOVLSOXRJP3ZCP4C","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"pith_short_8","alias_value":"YOVLSOXR","created_at":"2026-06-30T01:17:47Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:YOVLSOXRJP3ZCP4CM7EDR2IQYL","target":"record","payload":{"canonical_record":{"source":{"id":"2606.28955","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-27T14:53:03Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d39defa5e227f6d7d8b05c8d32215a837b1be0593b2f5f895d3f49d616ebd3fd","abstract_canon_sha256":"7712cb537c9dec24ad00735ac2044a5827105c31f6270c727fff42f273bc403e"},"schema_version":"1.0"},"canonical_sha256":"c3aab93af14bf7913f8267c838e910c2d024b99dbc4926f15a78d5f55146d01c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:17:47.099479Z","signature_b64":"Pjufg1oAohODtj03XJCEEHvIK0HDB/1Z8xjsyny3uHuMBUvZEx8KNvzY7RyV5OPKpU1dFMilzM1S4w+TtzNuBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c3aab93af14bf7913f8267c838e910c2d024b99dbc4926f15a78d5f55146d01c","last_reissued_at":"2026-06-30T01:17:47.098354Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:17:47.098354Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.28955","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-30T01:17:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DjepSDK+l5AYt+kGNZiMkVFGqkmy1PR7eCJoXGTRqWMhldacC2vbhfgpZA1xKvCBK7gztygdl9l8ZuxGs+RLCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-30T19:10:22.664805Z"},"content_sha256":"6b5f073e482de786c594ea4586318b403b4ec3927fc24f26b9cdec3abff243fa","schema_version":"1.0","event_id":"sha256:6b5f073e482de786c594ea4586318b403b4ec3927fc24f26b9cdec3abff243fa"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:YOVLSOXRJP3ZCP4CM7EDR2IQYL","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Modification-Considering Value Learning for Reward Hacking Mitigation in RL","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Evgenii Opryshko, Igor Gilitschenski, Umangi Jain","submitted_at":"2026-06-27T14:53:03Z","abstract_excerpt":"Reinforcement learning agents can exploit misspecified reward signals to achieve high apparent returns while failing on the intended objective, a failure mode known as reward hacking. Existing practical defenses typically constrain policy updates to stay near a known safe reference, creating a tension between suppressing hacking and permitting legitimate improvement. We propose Modification-Considering Value Learning (MCVL), which operationalizes the theoretical idea of current utility optimization for standard value-based RL. MCVL wraps an off-policy learner and treats each incoming transitio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.28955","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/2606.28955/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-30T01:17:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HuGEJeD7+RKH+LzPWtgR0Djb55t4GwDOYc1TZgtdYfGq+hcOAY3mlkL2by36GMS1Ye6LvfUba+vN1qqjfUkxBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-30T19:10:22.665418Z"},"content_sha256":"6f3a34e6f4d47717ed469e9d34ba2c7857353d2dfb3b5115c462dd4afa5bdda6","schema_version":"1.0","event_id":"sha256:6f3a34e6f4d47717ed469e9d34ba2c7857353d2dfb3b5115c462dd4afa5bdda6"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/YOVLSOXRJP3ZCP4CM7EDR2IQYL/bundle.json","state_url":"https://pith.science/pith/YOVLSOXRJP3ZCP4CM7EDR2IQYL/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/YOVLSOXRJP3ZCP4CM7EDR2IQYL/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-30T19:10:22Z","links":{"resolver":"https://pith.science/pith/YOVLSOXRJP3ZCP4CM7EDR2IQYL","bundle":"https://pith.science/pith/YOVLSOXRJP3ZCP4CM7EDR2IQYL/bundle.json","state":"https://pith.science/pith/YOVLSOXRJP3ZCP4CM7EDR2IQYL/state.json","well_known_bundle":"https://pith.science/.well-known/pith/YOVLSOXRJP3ZCP4CM7EDR2IQYL/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:YOVLSOXRJP3ZCP4CM7EDR2IQYL","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":"7712cb537c9dec24ad00735ac2044a5827105c31f6270c727fff42f273bc403e","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-27T14:53:03Z","title_canon_sha256":"d39defa5e227f6d7d8b05c8d32215a837b1be0593b2f5f895d3f49d616ebd3fd"},"schema_version":"1.0","source":{"id":"2606.28955","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.28955","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"arxiv_version","alias_value":"2606.28955v1","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.28955","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"pith_short_12","alias_value":"YOVLSOXRJP3Z","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"pith_short_16","alias_value":"YOVLSOXRJP3ZCP4C","created_at":"2026-06-30T01:17:47Z"},{"alias_kind":"pith_short_8","alias_value":"YOVLSOXR","created_at":"2026-06-30T01:17:47Z"}],"graph_snapshots":[{"event_id":"sha256:6f3a34e6f4d47717ed469e9d34ba2c7857353d2dfb3b5115c462dd4afa5bdda6","target":"graph","created_at":"2026-06-30T01:17:47Z","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/2606.28955/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement learning agents can exploit misspecified reward signals to achieve high apparent returns while failing on the intended objective, a failure mode known as reward hacking. Existing practical defenses typically constrain policy updates to stay near a known safe reference, creating a tension between suppressing hacking and permitting legitimate improvement. We propose Modification-Considering Value Learning (MCVL), which operationalizes the theoretical idea of current utility optimization for standard value-based RL. MCVL wraps an off-policy learner and treats each incoming transitio","authors_text":"Evgenii Opryshko, Igor Gilitschenski, Umangi Jain","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-27T14:53:03Z","title":"Modification-Considering Value Learning for Reward Hacking Mitigation in RL"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.28955","kind":"arxiv","version":1},"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":"sha256:6b5f073e482de786c594ea4586318b403b4ec3927fc24f26b9cdec3abff243fa","target":"record","created_at":"2026-06-30T01:17:47Z","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":"7712cb537c9dec24ad00735ac2044a5827105c31f6270c727fff42f273bc403e","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-27T14:53:03Z","title_canon_sha256":"d39defa5e227f6d7d8b05c8d32215a837b1be0593b2f5f895d3f49d616ebd3fd"},"schema_version":"1.0","source":{"id":"2606.28955","kind":"arxiv","version":1}},"canonical_sha256":"c3aab93af14bf7913f8267c838e910c2d024b99dbc4926f15a78d5f55146d01c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c3aab93af14bf7913f8267c838e910c2d024b99dbc4926f15a78d5f55146d01c","first_computed_at":"2026-06-30T01:17:47.098354Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-30T01:17:47.098354Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Pjufg1oAohODtj03XJCEEHvIK0HDB/1Z8xjsyny3uHuMBUvZEx8KNvzY7RyV5OPKpU1dFMilzM1S4w+TtzNuBA==","signature_status":"signed_v1","signed_at":"2026-06-30T01:17:47.099479Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.28955","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6b5f073e482de786c594ea4586318b403b4ec3927fc24f26b9cdec3abff243fa","sha256:6f3a34e6f4d47717ed469e9d34ba2c7857353d2dfb3b5115c462dd4afa5bdda6"],"state_sha256":"33f165d6d615f54033498ec0b45da5e6685c836226bf7d0e751faddd2759af7e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"k21Sa2g5UuHBucq7tQi1fvO59tTBn9DOiB/KnVhTEtU1J1MWbLbob03VjOLHTE9T9X9xNaTq53uYhQH5TdQlAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-30T19:10:22.668288Z","bundle_sha256":"3b4c23c02f5d6285b088b801a3c8cb01f47748b1a58966f3880aa6cea7c97cba"}}