{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:DZNJNSIGD6NQ4IVFKAVQMITQTJ","short_pith_number":"pith:DZNJNSIG","canonical_record":{"source":{"id":"2501.09620","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-01-16T16:00:37Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7e7d276700063c9a8be1681c6bb748fb753ba5942c0875e812154fc75209cbf7","abstract_canon_sha256":"31bdabe9602a1ec294beecf7d7f100007d03bcd3810cd95bad480ace2abc4152"},"schema_version":"1.0"},"canonical_sha256":"1e5a96c9061f9b0e22a5502b0622709a508d374cc9730ed1dbb0f877f9a29a6b","source":{"kind":"arxiv","id":"2501.09620","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.09620","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"arxiv_version","alias_value":"2501.09620v2","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.09620","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"pith_short_12","alias_value":"DZNJNSIGD6NQ","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"pith_short_16","alias_value":"DZNJNSIGD6NQ4IVF","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"pith_short_8","alias_value":"DZNJNSIG","created_at":"2026-07-05T11:11:39Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:DZNJNSIGD6NQ4IVFKAVQMITQTJ","target":"record","payload":{"canonical_record":{"source":{"id":"2501.09620","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-01-16T16:00:37Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7e7d276700063c9a8be1681c6bb748fb753ba5942c0875e812154fc75209cbf7","abstract_canon_sha256":"31bdabe9602a1ec294beecf7d7f100007d03bcd3810cd95bad480ace2abc4152"},"schema_version":"1.0"},"canonical_sha256":"1e5a96c9061f9b0e22a5502b0622709a508d374cc9730ed1dbb0f877f9a29a6b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:11:39.807555Z","signature_b64":"8Hcc4ABLD3+AsuJWIIOQDNrqYkr1+jUvC4HmhYCoKes0/PzyaZzW+jd1TiBJTyHCLXM2mnSkmqRkcal7K5BkCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1e5a96c9061f9b0e22a5502b0622709a508d374cc9730ed1dbb0f877f9a29a6b","last_reissued_at":"2026-07-05T11:11:39.807014Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:11:39.807014Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.09620","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":"2026-07-05T11:11:39Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"074MvyiUbwx/wW5sFpj/kJQfUTUPbZqQf4cJQt/Ggl3WwkTXGf4v+/BzNhZwQiwCu7j9/k09tts1Ga7uxB8jBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-01T21:03:58.867690Z"},"content_sha256":"86b9a884a25017bb493cbd6960e0f724ff4ef303a8ba4f272b4839d959920f51","schema_version":"1.0","event_id":"sha256:86b9a884a25017bb493cbd6960e0f724ff4ef303a8ba4f272b4839d959920f51"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:DZNJNSIGD6NQ4IVFKAVQMITQTJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Beyond Reward Hacking: Causal Rewards for Large Language Model Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chaoqi Wang, Chen Zhu, Hao Ma, Jiayi Liu, Lizhu Zhang, Sinong Wang, Xiangjun Fan, Yibo Jiang, Yuxin Chen, Zhaorun Chen, Zhuokai Zhao","submitted_at":"2025-01-16T16:00:37Z","abstract_excerpt":"Recent advances in large language models (LLMs) have demonstrated significant progress in performing complex tasks. While Reinforcement Learning from Human Feedback (RLHF) has been effective in aligning LLMs with human preferences, it is susceptible to spurious correlations in reward modeling. Consequently, it often introduces biases-such as length bias, sycophancy, conceptual bias, and discrimination-that hinder the model's ability to capture true causal relationships. To address this, we propose a novel causal reward modeling approach that integrates causality to mitigate these spurious corr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.09620","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/2501.09620/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":"2026-07-05T11:11:39Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"yL1GpWUyIu1A1U4wBbg91Qw8R9JaqHT3UBArjVomI3nqGY5CNo95JaWUpDfVszb7gnh1HdtF48Z7E6bVinN7Dw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-01T21:03:58.868233Z"},"content_sha256":"ad0b8b37b95fbfad0f0639894c2ecce2c06f68c5ce52c9e7d669561451b33443","schema_version":"1.0","event_id":"sha256:ad0b8b37b95fbfad0f0639894c2ecce2c06f68c5ce52c9e7d669561451b33443"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DZNJNSIGD6NQ4IVFKAVQMITQTJ/bundle.json","state_url":"https://pith.science/pith/DZNJNSIGD6NQ4IVFKAVQMITQTJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DZNJNSIGD6NQ4IVFKAVQMITQTJ/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":"2026-08-01T21:03:58Z","links":{"resolver":"https://pith.science/pith/DZNJNSIGD6NQ4IVFKAVQMITQTJ","bundle":"https://pith.science/pith/DZNJNSIGD6NQ4IVFKAVQMITQTJ/bundle.json","state":"https://pith.science/pith/DZNJNSIGD6NQ4IVFKAVQMITQTJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DZNJNSIGD6NQ4IVFKAVQMITQTJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:DZNJNSIGD6NQ4IVFKAVQMITQTJ","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":"31bdabe9602a1ec294beecf7d7f100007d03bcd3810cd95bad480ace2abc4152","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-01-16T16:00:37Z","title_canon_sha256":"7e7d276700063c9a8be1681c6bb748fb753ba5942c0875e812154fc75209cbf7"},"schema_version":"1.0","source":{"id":"2501.09620","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.09620","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"arxiv_version","alias_value":"2501.09620v2","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.09620","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"pith_short_12","alias_value":"DZNJNSIGD6NQ","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"pith_short_16","alias_value":"DZNJNSIGD6NQ4IVF","created_at":"2026-07-05T11:11:39Z"},{"alias_kind":"pith_short_8","alias_value":"DZNJNSIG","created_at":"2026-07-05T11:11:39Z"}],"graph_snapshots":[{"event_id":"sha256:ad0b8b37b95fbfad0f0639894c2ecce2c06f68c5ce52c9e7d669561451b33443","target":"graph","created_at":"2026-07-05T11:11:39Z","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/2501.09620/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent advances in large language models (LLMs) have demonstrated significant progress in performing complex tasks. While Reinforcement Learning from Human Feedback (RLHF) has been effective in aligning LLMs with human preferences, it is susceptible to spurious correlations in reward modeling. Consequently, it often introduces biases-such as length bias, sycophancy, conceptual bias, and discrimination-that hinder the model's ability to capture true causal relationships. To address this, we propose a novel causal reward modeling approach that integrates causality to mitigate these spurious corr","authors_text":"Chaoqi Wang, Chen Zhu, Hao Ma, Jiayi Liu, Lizhu Zhang, Sinong Wang, Xiangjun Fan, Yibo Jiang, Yuxin Chen, Zhaorun Chen, Zhuokai Zhao","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-01-16T16:00:37Z","title":"Beyond Reward Hacking: Causal Rewards for Large Language Model Alignment"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.09620","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":"sha256:86b9a884a25017bb493cbd6960e0f724ff4ef303a8ba4f272b4839d959920f51","target":"record","created_at":"2026-07-05T11:11:39Z","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":"31bdabe9602a1ec294beecf7d7f100007d03bcd3810cd95bad480ace2abc4152","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-01-16T16:00:37Z","title_canon_sha256":"7e7d276700063c9a8be1681c6bb748fb753ba5942c0875e812154fc75209cbf7"},"schema_version":"1.0","source":{"id":"2501.09620","kind":"arxiv","version":2}},"canonical_sha256":"1e5a96c9061f9b0e22a5502b0622709a508d374cc9730ed1dbb0f877f9a29a6b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1e5a96c9061f9b0e22a5502b0622709a508d374cc9730ed1dbb0f877f9a29a6b","first_computed_at":"2026-07-05T11:11:39.807014Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:11:39.807014Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"8Hcc4ABLD3+AsuJWIIOQDNrqYkr1+jUvC4HmhYCoKes0/PzyaZzW+jd1TiBJTyHCLXM2mnSkmqRkcal7K5BkCw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:11:39.807555Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.09620","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:86b9a884a25017bb493cbd6960e0f724ff4ef303a8ba4f272b4839d959920f51","sha256:ad0b8b37b95fbfad0f0639894c2ecce2c06f68c5ce52c9e7d669561451b33443"],"state_sha256":"44c9969306ce15c5bf0d5d246f269250fcdee4baff4ce713b0fa0b153ec062f6"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"F9k8hfpn9v/ZJ/E1YqxhzMENb+RDfGonKDruBSS7cRQ30uI1ifnnUH9djmRVxDseeIgYdI94fZP2GuwDI351Dg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-01T21:03:58.874021Z","bundle_sha256":"5e03e050c8c56feee1e8e00596f263b834604d5de04f141829715252bb59399d"}}