{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:ASN56S5CBHTSR7MCMB5X7OPDQL","short_pith_number":"pith:ASN56S5C","canonical_record":{"source":{"id":"2406.10216","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-14T17:49:59Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cb7d709f77324a677b0269d5af199b4e948ea13c15ebad5666676a21ad616ea0","abstract_canon_sha256":"62db7e74d90b88a4425c3ce4b9d4df76b62e0418111886c859bbfb35cbaa4670"},"schema_version":"1.0"},"canonical_sha256":"049bdf4ba209e728fd82607b7fb9e382e301681bef2cf3ffe1b0a86c1e40e009","source":{"kind":"arxiv","id":"2406.10216","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.10216","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"arxiv_version","alias_value":"2406.10216v2","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.10216","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"pith_short_12","alias_value":"ASN56S5CBHTS","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"pith_short_16","alias_value":"ASN56S5CBHTSR7MC","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"pith_short_8","alias_value":"ASN56S5C","created_at":"2026-07-05T09:24:36Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:ASN56S5CBHTSR7MCMB5X7OPDQL","target":"record","payload":{"canonical_record":{"source":{"id":"2406.10216","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-14T17:49:59Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cb7d709f77324a677b0269d5af199b4e948ea13c15ebad5666676a21ad616ea0","abstract_canon_sha256":"62db7e74d90b88a4425c3ce4b9d4df76b62e0418111886c859bbfb35cbaa4670"},"schema_version":"1.0"},"canonical_sha256":"049bdf4ba209e728fd82607b7fb9e382e301681bef2cf3ffe1b0a86c1e40e009","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:24:36.088004Z","signature_b64":"7RcHtdhMWn4Sq/Jww2N4KrhiCPpJZlq29Q9sSPlHdIA0fKuG5HD6JXI1brxsFYxEQiCQTJLrftN/Yr/iKC22Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"049bdf4ba209e728fd82607b7fb9e382e301681bef2cf3ffe1b0a86c1e40e009","last_reissued_at":"2026-07-05T09:24:36.087497Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:24:36.087497Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.10216","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:24:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lWl0Q7WyJsT0O+Nok8bpHGyR7pkBMt8M3mD7YlbycS8yfRLiIhCNMrp99VWUo4ZWk6xgfm2dboae9EjugD/DCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T12:03:14.531697Z"},"content_sha256":"39d6fdcfb01f11564535132df2d418136bbc8290812f809235bb4c67e7518b30","schema_version":"1.0","event_id":"sha256:39d6fdcfb01f11564535132df2d418136bbc8290812f809235bb4c67e7518b30"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:ASN56S5CBHTSR7MCMB5X7OPDQL","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Huan Zhang, Rui Yang, Ruomeng Ding, Tong Zhang, Yong Lin","submitted_at":"2024-06-14T17:49:59Z","abstract_excerpt":"Reward models trained on human preference data have been proven to effectively align Large Language Models (LLMs) with human intent within the framework of reinforcement learning from human feedback (RLHF). However, current reward models have limited generalization capabilities to unseen prompts and responses, which can lead to an unexpected phenomenon known as reward over-optimization, resulting in a decline in actual performance due to excessive optimization of rewards. While previous research has advocated for constraining policy optimization, our study introduces a novel approach to enhanc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.10216","kind":"arxiv","version":2},"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/2406.10216/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":"2026-07-05T09:24:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"nvCmC6ZxpNZCTvmi0KOKNQNpyDBvaxe1PiOKoLnHVRLBF6jt0HYRtLg0GRN4xjH+in2oleGV6zc/k+ddyEoOBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T12:03:14.532203Z"},"content_sha256":"885024927185ed865c7b1e0e5b542f1589367a772f635c8896aa8a50e5658f19","schema_version":"1.0","event_id":"sha256:885024927185ed865c7b1e0e5b542f1589367a772f635c8896aa8a50e5658f19"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ASN56S5CBHTSR7MCMB5X7OPDQL/bundle.json","state_url":"https://pith.science/pith/ASN56S5CBHTSR7MCMB5X7OPDQL/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ASN56S5CBHTSR7MCMB5X7OPDQL/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":"2026-08-08T12:03:14Z","links":{"resolver":"https://pith.science/pith/ASN56S5CBHTSR7MCMB5X7OPDQL","bundle":"https://pith.science/pith/ASN56S5CBHTSR7MCMB5X7OPDQL/bundle.json","state":"https://pith.science/pith/ASN56S5CBHTSR7MCMB5X7OPDQL/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ASN56S5CBHTSR7MCMB5X7OPDQL/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:ASN56S5CBHTSR7MCMB5X7OPDQL","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":"62db7e74d90b88a4425c3ce4b9d4df76b62e0418111886c859bbfb35cbaa4670","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-14T17:49:59Z","title_canon_sha256":"cb7d709f77324a677b0269d5af199b4e948ea13c15ebad5666676a21ad616ea0"},"schema_version":"1.0","source":{"id":"2406.10216","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.10216","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"arxiv_version","alias_value":"2406.10216v2","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.10216","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"pith_short_12","alias_value":"ASN56S5CBHTS","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"pith_short_16","alias_value":"ASN56S5CBHTSR7MC","created_at":"2026-07-05T09:24:36Z"},{"alias_kind":"pith_short_8","alias_value":"ASN56S5C","created_at":"2026-07-05T09:24:36Z"}],"graph_snapshots":[{"event_id":"sha256:885024927185ed865c7b1e0e5b542f1589367a772f635c8896aa8a50e5658f19","target":"graph","created_at":"2026-07-05T09:24:36Z","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/2406.10216/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reward models trained on human preference data have been proven to effectively align Large Language Models (LLMs) with human intent within the framework of reinforcement learning from human feedback (RLHF). However, current reward models have limited generalization capabilities to unseen prompts and responses, which can lead to an unexpected phenomenon known as reward over-optimization, resulting in a decline in actual performance due to excessive optimization of rewards. While previous research has advocated for constraining policy optimization, our study introduces a novel approach to enhanc","authors_text":"Huan Zhang, Rui Yang, Ruomeng Ding, Tong Zhang, Yong Lin","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-14T17:49:59Z","title":"Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.10216","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":"sha256:39d6fdcfb01f11564535132df2d418136bbc8290812f809235bb4c67e7518b30","target":"record","created_at":"2026-07-05T09:24:36Z","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":"62db7e74d90b88a4425c3ce4b9d4df76b62e0418111886c859bbfb35cbaa4670","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-14T17:49:59Z","title_canon_sha256":"cb7d709f77324a677b0269d5af199b4e948ea13c15ebad5666676a21ad616ea0"},"schema_version":"1.0","source":{"id":"2406.10216","kind":"arxiv","version":2}},"canonical_sha256":"049bdf4ba209e728fd82607b7fb9e382e301681bef2cf3ffe1b0a86c1e40e009","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"049bdf4ba209e728fd82607b7fb9e382e301681bef2cf3ffe1b0a86c1e40e009","first_computed_at":"2026-07-05T09:24:36.087497Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:24:36.087497Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"7RcHtdhMWn4Sq/Jww2N4KrhiCPpJZlq29Q9sSPlHdIA0fKuG5HD6JXI1brxsFYxEQiCQTJLrftN/Yr/iKC22Cg==","signature_status":"signed_v1","signed_at":"2026-07-05T09:24:36.088004Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.10216","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:39d6fdcfb01f11564535132df2d418136bbc8290812f809235bb4c67e7518b30","sha256:885024927185ed865c7b1e0e5b542f1589367a772f635c8896aa8a50e5658f19"],"state_sha256":"89c168258c46334f60ab28bdfb240526c19731c6be51e1fbe8949734ae9ea78f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"yCeLb/F93KEf7mFOcs8O6TYdMVK0Vhd7mibgcMp61bYai7IsYy9yq267mEihPde54qJBSFei2t/lQ54j7S3JCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T12:03:14.536703Z","bundle_sha256":"3c8c3ae42d99364a87a8a7b3396802dec2e1f15d51ad7c56066793f977bb4d3d"}}