{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:FCVT3PGQWPQE6TUORDWKHQRGOR","short_pith_number":"pith:FCVT3PGQ","canonical_record":{"source":{"id":"2606.09073","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T06:15:30Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"451827b14a685e8356f6272542177785b23095c6ef77b1b4a8df41aed4d801aa","abstract_canon_sha256":"139ce02a7345aad79683aba8ceaefe81cc4896eac113a4d97f191876776fb5c2"},"schema_version":"1.0"},"canonical_sha256":"28ab3dbcd0b3e04f4e8e88eca3c226747cc5d4b8663698c56a631bb18aaf620a","source":{"kind":"arxiv","id":"2606.09073","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.09073","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"arxiv_version","alias_value":"2606.09073v1","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09073","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"pith_short_12","alias_value":"FCVT3PGQWPQE","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"pith_short_16","alias_value":"FCVT3PGQWPQE6TUO","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"pith_short_8","alias_value":"FCVT3PGQ","created_at":"2026-06-09T02:07:57Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:FCVT3PGQWPQE6TUORDWKHQRGOR","target":"record","payload":{"canonical_record":{"source":{"id":"2606.09073","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T06:15:30Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"451827b14a685e8356f6272542177785b23095c6ef77b1b4a8df41aed4d801aa","abstract_canon_sha256":"139ce02a7345aad79683aba8ceaefe81cc4896eac113a4d97f191876776fb5c2"},"schema_version":"1.0"},"canonical_sha256":"28ab3dbcd0b3e04f4e8e88eca3c226747cc5d4b8663698c56a631bb18aaf620a","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:07:57.176823Z","signature_b64":"NP2sKM2/gD1R+qRoi7XunL5No3kDoBhYsOVdR51RN93hanrimYoI3K25QdRTz8L3X3+1rkaVtvQXDO/ZJbj0Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"28ab3dbcd0b3e04f4e8e88eca3c226747cc5d4b8663698c56a631bb18aaf620a","last_reissued_at":"2026-06-09T02:07:57.176264Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:07:57.176264Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.09073","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":"2026-06-09T02:07:57Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"nLK9BFc4q5S1MKkKDVNjrTFZYu3ULR39IRbsOWOVTqmkHfhhGAP+DoPM9YuR6KAfY9Ebb28Gm/iYSyTmhRtqDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-03T08:16:42.914915Z"},"content_sha256":"0d1304f8ea1602c0847113a21d9e153d609aec3e79e5630c9abd6a5dd9fc26a6","schema_version":"1.0","event_id":"sha256:0d1304f8ea1602c0847113a21d9e153d609aec3e79e5630c9abd6a5dd9fc26a6"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:FCVT3PGQWPQE6TUORDWKHQRGOR","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"A Unifying Lens on Reward Uncertainty in RLHF","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ely Hahami, Jack Benarroch Jedlicki, Ray Zhou, Yoel Zimmermann","submitted_at":"2026-06-08T06:15:30Z","abstract_excerpt":"Reinforcement learning from human feedback (RLHF) is bottlenecked by \\emph{reward hacking}, where the policy exploits errors in a proxy reward model (RM) and produces high RM scores without genuine quality gains. A natural mitigation is \\emph{pessimism}: penalizing rewards in regions where the RM is uncertain. However, standard scalar RMs provide no principled notion of uncertainty. We argue that the right object is a \\emph{distributional} reward model $p(r\\mid x,y)$. Under either a Bayesian inference or a KL-distributionally robust optimization (KL-DRO) lens, the KL-regularized RLHF objective"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09073","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/2606.09073/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":"2026-06-09T02:07:57Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tXnTGYwpSLBFm3pgmpTdMjf9+6o+jJXzmHRUJIniBF6BO2D/8JWdeWPwjL53PXfxrhQoZBaUO207ZBuAlCx0CA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-03T08:16:42.915293Z"},"content_sha256":"055fa8ad376bdea6c861bc7303a5f0b5e7fa8cc051bda34e6ac36cfa661945b1","schema_version":"1.0","event_id":"sha256:055fa8ad376bdea6c861bc7303a5f0b5e7fa8cc051bda34e6ac36cfa661945b1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FCVT3PGQWPQE6TUORDWKHQRGOR/bundle.json","state_url":"https://pith.science/pith/FCVT3PGQWPQE6TUORDWKHQRGOR/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FCVT3PGQWPQE6TUORDWKHQRGOR/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":"2026-07-03T08:16:42Z","links":{"resolver":"https://pith.science/pith/FCVT3PGQWPQE6TUORDWKHQRGOR","bundle":"https://pith.science/pith/FCVT3PGQWPQE6TUORDWKHQRGOR/bundle.json","state":"https://pith.science/pith/FCVT3PGQWPQE6TUORDWKHQRGOR/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FCVT3PGQWPQE6TUORDWKHQRGOR/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:FCVT3PGQWPQE6TUORDWKHQRGOR","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":"139ce02a7345aad79683aba8ceaefe81cc4896eac113a4d97f191876776fb5c2","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T06:15:30Z","title_canon_sha256":"451827b14a685e8356f6272542177785b23095c6ef77b1b4a8df41aed4d801aa"},"schema_version":"1.0","source":{"id":"2606.09073","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.09073","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"arxiv_version","alias_value":"2606.09073v1","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09073","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"pith_short_12","alias_value":"FCVT3PGQWPQE","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"pith_short_16","alias_value":"FCVT3PGQWPQE6TUO","created_at":"2026-06-09T02:07:57Z"},{"alias_kind":"pith_short_8","alias_value":"FCVT3PGQ","created_at":"2026-06-09T02:07:57Z"}],"graph_snapshots":[{"event_id":"sha256:055fa8ad376bdea6c861bc7303a5f0b5e7fa8cc051bda34e6ac36cfa661945b1","target":"graph","created_at":"2026-06-09T02:07:57Z","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/2606.09073/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement learning from human feedback (RLHF) is bottlenecked by \\emph{reward hacking}, where the policy exploits errors in a proxy reward model (RM) and produces high RM scores without genuine quality gains. A natural mitigation is \\emph{pessimism}: penalizing rewards in regions where the RM is uncertain. However, standard scalar RMs provide no principled notion of uncertainty. We argue that the right object is a \\emph{distributional} reward model $p(r\\mid x,y)$. Under either a Bayesian inference or a KL-distributionally robust optimization (KL-DRO) lens, the KL-regularized RLHF objective","authors_text":"Ely Hahami, Jack Benarroch Jedlicki, Ray Zhou, Yoel Zimmermann","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T06:15:30Z","title":"A Unifying Lens on Reward Uncertainty in RLHF"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09073","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":"sha256:0d1304f8ea1602c0847113a21d9e153d609aec3e79e5630c9abd6a5dd9fc26a6","target":"record","created_at":"2026-06-09T02:07:57Z","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":"139ce02a7345aad79683aba8ceaefe81cc4896eac113a4d97f191876776fb5c2","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T06:15:30Z","title_canon_sha256":"451827b14a685e8356f6272542177785b23095c6ef77b1b4a8df41aed4d801aa"},"schema_version":"1.0","source":{"id":"2606.09073","kind":"arxiv","version":1}},"canonical_sha256":"28ab3dbcd0b3e04f4e8e88eca3c226747cc5d4b8663698c56a631bb18aaf620a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"28ab3dbcd0b3e04f4e8e88eca3c226747cc5d4b8663698c56a631bb18aaf620a","first_computed_at":"2026-06-09T02:07:57.176264Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-09T02:07:57.176264Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"NP2sKM2/gD1R+qRoi7XunL5No3kDoBhYsOVdR51RN93hanrimYoI3K25QdRTz8L3X3+1rkaVtvQXDO/ZJbj0Bw==","signature_status":"signed_v1","signed_at":"2026-06-09T02:07:57.176823Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.09073","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0d1304f8ea1602c0847113a21d9e153d609aec3e79e5630c9abd6a5dd9fc26a6","sha256:055fa8ad376bdea6c861bc7303a5f0b5e7fa8cc051bda34e6ac36cfa661945b1"],"state_sha256":"1aea1ed0ec111f61cde2ef0d3ade105cd6f377f7697d1f56f9f597fd6c80592f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rV2nKzgB7fk4eoDTz12QClSjDZ+1zTTau+sxcgIUJj65qnpKig9fezS1ymy93//O6+mHQ6+ZAgb+6TJdawePDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-03T08:16:42.917309Z","bundle_sha256":"8a40719d50db3febd02b65e15db936f43870c5864ca59e1d4e6a5552a6e574e6"}}