{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:LUVWPRXPUKMJPVGMRWNS5KGF6S","short_pith_number":"pith:LUVWPRXP","canonical_record":{"source":{"id":"2502.14619","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T14:57:14Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"086b0e3743a763e5e626525a24ffef16c1b30aad8ab1fb21f426fe326f388ab5","abstract_canon_sha256":"fbf0ad564924daacd32cb63452a707797055a54cbefa09477cce67388c4ef28b"},"schema_version":"1.0"},"canonical_sha256":"5d2b67c6efa29897d4cc8d9b2ea8c5f48597324d3abf2e7cbc30b8ed7d1c2999","source":{"kind":"arxiv","id":"2502.14619","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.14619","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"arxiv_version","alias_value":"2502.14619v1","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.14619","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"pith_short_12","alias_value":"LUVWPRXPUKMJ","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"pith_short_16","alias_value":"LUVWPRXPUKMJPVGM","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"pith_short_8","alias_value":"LUVWPRXP","created_at":"2026-07-05T10:17:30Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:LUVWPRXPUKMJPVGMRWNS5KGF6S","target":"record","payload":{"canonical_record":{"source":{"id":"2502.14619","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T14:57:14Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"086b0e3743a763e5e626525a24ffef16c1b30aad8ab1fb21f426fe326f388ab5","abstract_canon_sha256":"fbf0ad564924daacd32cb63452a707797055a54cbefa09477cce67388c4ef28b"},"schema_version":"1.0"},"canonical_sha256":"5d2b67c6efa29897d4cc8d9b2ea8c5f48597324d3abf2e7cbc30b8ed7d1c2999","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:17:30.174357Z","signature_b64":"p/OFxrmDvU0mwx7IKiOVGgnEXv+ce25aaaSxfSbW57XavLksbFtLP4aw3EY3KLhpFuEfsUGKiMJZeYtaFiAwBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5d2b67c6efa29897d4cc8d9b2ea8c5f48597324d3abf2e7cbc30b8ed7d1c2999","last_reissued_at":"2026-07-05T10:17:30.173757Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:17:30.173757Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.14619","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":"2026-07-05T10:17:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1zUNZni20LledIhixBDdw/vrdEyYkDjRrCwXUcXSzsqwpIb6ryWE0SJZbs1xp1MZuOEckTUppjTdwLZh3ffcBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T12:33:01.805897Z"},"content_sha256":"7bf5ced6d5d98704e6e7da06fe9c6061e4eac2332bd5c21f547783e6a82be14f","schema_version":"1.0","event_id":"sha256:7bf5ced6d5d98704e6e7da06fe9c6061e4eac2332bd5c21f547783e6a82be14f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:LUVWPRXPUKMJPVGMRWNS5KGF6S","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Reward Models Identify Consistency, Not Causality","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Caiming Xiong, HanZe Dong, Junnan Li, Lei Wang, Yuhui Xu","submitted_at":"2025-02-20T14:57:14Z","abstract_excerpt":"Reward models (RMs) play a crucial role in aligning large language models (LLMs) with human preferences and enhancing reasoning quality. Traditionally, RMs are trained to rank candidate outputs based on their correctness and coherence. However, in this work, we present several surprising findings that challenge common assumptions about RM behavior. Our analysis reveals that state-of-the-art reward models prioritize structural consistency over causal correctness. Specifically, removing the problem statement has minimal impact on reward scores, whereas altering numerical values or disrupting the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.14619","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/2502.14619/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":"2026-07-05T10:17:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0Pes44u2hpdzclufQ/5Lz382Y+HptOcuILuw7PwQUqfkHIh2dVm2LLdXR3yE6TO2kdD6RMdAGp64P3Zh3DYIAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T12:33:01.806186Z"},"content_sha256":"cc31a4037bbabb010b652fff2b58a04f83eb9354c1b74d4a6ebec8c48ec7bf5c","schema_version":"1.0","event_id":"sha256:cc31a4037bbabb010b652fff2b58a04f83eb9354c1b74d4a6ebec8c48ec7bf5c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/LUVWPRXPUKMJPVGMRWNS5KGF6S/bundle.json","state_url":"https://pith.science/pith/LUVWPRXPUKMJPVGMRWNS5KGF6S/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/LUVWPRXPUKMJPVGMRWNS5KGF6S/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":"2026-08-05T12:33:01Z","links":{"resolver":"https://pith.science/pith/LUVWPRXPUKMJPVGMRWNS5KGF6S","bundle":"https://pith.science/pith/LUVWPRXPUKMJPVGMRWNS5KGF6S/bundle.json","state":"https://pith.science/pith/LUVWPRXPUKMJPVGMRWNS5KGF6S/state.json","well_known_bundle":"https://pith.science/.well-known/pith/LUVWPRXPUKMJPVGMRWNS5KGF6S/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:LUVWPRXPUKMJPVGMRWNS5KGF6S","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":"fbf0ad564924daacd32cb63452a707797055a54cbefa09477cce67388c4ef28b","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T14:57:14Z","title_canon_sha256":"086b0e3743a763e5e626525a24ffef16c1b30aad8ab1fb21f426fe326f388ab5"},"schema_version":"1.0","source":{"id":"2502.14619","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.14619","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"arxiv_version","alias_value":"2502.14619v1","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.14619","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"pith_short_12","alias_value":"LUVWPRXPUKMJ","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"pith_short_16","alias_value":"LUVWPRXPUKMJPVGM","created_at":"2026-07-05T10:17:30Z"},{"alias_kind":"pith_short_8","alias_value":"LUVWPRXP","created_at":"2026-07-05T10:17:30Z"}],"graph_snapshots":[{"event_id":"sha256:cc31a4037bbabb010b652fff2b58a04f83eb9354c1b74d4a6ebec8c48ec7bf5c","target":"graph","created_at":"2026-07-05T10:17:30Z","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/2502.14619/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reward models (RMs) play a crucial role in aligning large language models (LLMs) with human preferences and enhancing reasoning quality. Traditionally, RMs are trained to rank candidate outputs based on their correctness and coherence. However, in this work, we present several surprising findings that challenge common assumptions about RM behavior. Our analysis reveals that state-of-the-art reward models prioritize structural consistency over causal correctness. Specifically, removing the problem statement has minimal impact on reward scores, whereas altering numerical values or disrupting the","authors_text":"Caiming Xiong, HanZe Dong, Junnan Li, Lei Wang, Yuhui Xu","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T14:57:14Z","title":"Reward Models Identify Consistency, Not Causality"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.14619","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":"sha256:7bf5ced6d5d98704e6e7da06fe9c6061e4eac2332bd5c21f547783e6a82be14f","target":"record","created_at":"2026-07-05T10:17:30Z","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":"fbf0ad564924daacd32cb63452a707797055a54cbefa09477cce67388c4ef28b","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T14:57:14Z","title_canon_sha256":"086b0e3743a763e5e626525a24ffef16c1b30aad8ab1fb21f426fe326f388ab5"},"schema_version":"1.0","source":{"id":"2502.14619","kind":"arxiv","version":1}},"canonical_sha256":"5d2b67c6efa29897d4cc8d9b2ea8c5f48597324d3abf2e7cbc30b8ed7d1c2999","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"5d2b67c6efa29897d4cc8d9b2ea8c5f48597324d3abf2e7cbc30b8ed7d1c2999","first_computed_at":"2026-07-05T10:17:30.173757Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:17:30.173757Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"p/OFxrmDvU0mwx7IKiOVGgnEXv+ce25aaaSxfSbW57XavLksbFtLP4aw3EY3KLhpFuEfsUGKiMJZeYtaFiAwBA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:17:30.174357Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.14619","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7bf5ced6d5d98704e6e7da06fe9c6061e4eac2332bd5c21f547783e6a82be14f","sha256:cc31a4037bbabb010b652fff2b58a04f83eb9354c1b74d4a6ebec8c48ec7bf5c"],"state_sha256":"15b2f4018260c196f93b812ea697a97437c3cebf57bc1739e0caafe7331a27f0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LdFS/bAFE3/zQkcjAPTHRqzVaB62kFH+G+abYSzBLZCGxDiTmXwfeNu5B7lwtTY6U9p0E3FpqzW0cOG7RUHuCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T12:33:01.809633Z","bundle_sha256":"2efaf8a8821906bc25a1573b2620b8af427c42e23e7b786a5dd657d677499094"}}