{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:CLYJY6HJJRC6DTRTGJIYIFV7B2","short_pith_number":"pith:CLYJY6HJ","canonical_record":{"source":{"id":"2411.16502","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-25T15:37:27Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"18c575c81f91d3b35f17629e4a976fcf48b09af6e1836ed8739637510a964047","abstract_canon_sha256":"613b2d784c3df2681a483e836939e98e171c891045bb82822ee2ce38208f6818"},"schema_version":"1.0"},"canonical_sha256":"12f09c78e94c45e1ce3332518416bf0e9d407c6a3db5ac98743d5aecea5a6a57","source":{"kind":"arxiv","id":"2411.16502","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2411.16502","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"arxiv_version","alias_value":"2411.16502v2","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.16502","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"pith_short_12","alias_value":"CLYJY6HJJRC6","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"pith_short_16","alias_value":"CLYJY6HJJRC6DTRT","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"pith_short_8","alias_value":"CLYJY6HJ","created_at":"2026-07-05T10:20:17Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:CLYJY6HJJRC6DTRTGJIYIFV7B2","target":"record","payload":{"canonical_record":{"source":{"id":"2411.16502","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-25T15:37:27Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"18c575c81f91d3b35f17629e4a976fcf48b09af6e1836ed8739637510a964047","abstract_canon_sha256":"613b2d784c3df2681a483e836939e98e171c891045bb82822ee2ce38208f6818"},"schema_version":"1.0"},"canonical_sha256":"12f09c78e94c45e1ce3332518416bf0e9d407c6a3db5ac98743d5aecea5a6a57","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:20:17.749871Z","signature_b64":"qnYU27NgYoDVqGkrLOe6nUNVx8QMeknHtA9qCPrWryG0hvTTfRQe/KrstSCceONiVY0BAwDXMnWI8M9YhcoKCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"12f09c78e94c45e1ce3332518416bf0e9d407c6a3db5ac98743d5aecea5a6a57","last_reissued_at":"2026-07-05T10:20:17.749282Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:20:17.749282Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2411.16502","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":"2026-07-05T10:20:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"axNfqQ0CgVKSpuLdWkLKebuzWVavRy0ZuJFNjmQ3IGVOvQuO2r4uXCBv0bq28cGlGvJhC2P1LN9OCwLQ6I25AQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T17:20:15.371814Z"},"content_sha256":"dc9502ba3645b936c709ed18d3a9f979a0be10aacf2e8ca3a34382df477ebabb","schema_version":"1.0","event_id":"sha256:dc9502ba3645b936c709ed18d3a9f979a0be10aacf2e8ca3a34382df477ebabb"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:CLYJY6HJJRC6DTRTGJIYIFV7B2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Interpreting Language Reward Models via Contrastive Explanations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Freddy Lecue, Junqi Jiang, Manuela Veloso, Saumitra Mishra, Tom Bewley","submitted_at":"2024-11-25T15:37:27Z","abstract_excerpt":"Reward models (RMs) are a crucial component in the alignment of large language models' (LLMs) outputs with human values. RMs approximate human preferences over possible LLM responses to the same prompt by predicting and comparing reward scores. However, as they are typically modified versions of LLMs with scalar output heads, RMs are large black boxes whose predictions are not explainable. More transparent RMs would enable improved trust in the alignment of LLMs. In this work, we propose to use contrastive explanations to explain any binary response comparison made by an RM. Specifically, we g"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.16502","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/2411.16502/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":"2026-07-05T10:20:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"CQKZqga40yxi85OQekTGAMoLW+q0tXGs3+nbNsLdPZMVRtc4jZDn/iDIM8MRNwdCe61SnrIAbfih1oy7laWhBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T17:20:15.372403Z"},"content_sha256":"f1fea8d59b0bed316b048abad6ced4c6d275759c10748857b13a8bab9eb5d0d2","schema_version":"1.0","event_id":"sha256:f1fea8d59b0bed316b048abad6ced4c6d275759c10748857b13a8bab9eb5d0d2"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/CLYJY6HJJRC6DTRTGJIYIFV7B2/bundle.json","state_url":"https://pith.science/pith/CLYJY6HJJRC6DTRTGJIYIFV7B2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/CLYJY6HJJRC6DTRTGJIYIFV7B2/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":"2026-08-09T17:20:15Z","links":{"resolver":"https://pith.science/pith/CLYJY6HJJRC6DTRTGJIYIFV7B2","bundle":"https://pith.science/pith/CLYJY6HJJRC6DTRTGJIYIFV7B2/bundle.json","state":"https://pith.science/pith/CLYJY6HJJRC6DTRTGJIYIFV7B2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/CLYJY6HJJRC6DTRTGJIYIFV7B2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:CLYJY6HJJRC6DTRTGJIYIFV7B2","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":"613b2d784c3df2681a483e836939e98e171c891045bb82822ee2ce38208f6818","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-25T15:37:27Z","title_canon_sha256":"18c575c81f91d3b35f17629e4a976fcf48b09af6e1836ed8739637510a964047"},"schema_version":"1.0","source":{"id":"2411.16502","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2411.16502","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"arxiv_version","alias_value":"2411.16502v2","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.16502","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"pith_short_12","alias_value":"CLYJY6HJJRC6","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"pith_short_16","alias_value":"CLYJY6HJJRC6DTRT","created_at":"2026-07-05T10:20:17Z"},{"alias_kind":"pith_short_8","alias_value":"CLYJY6HJ","created_at":"2026-07-05T10:20:17Z"}],"graph_snapshots":[{"event_id":"sha256:f1fea8d59b0bed316b048abad6ced4c6d275759c10748857b13a8bab9eb5d0d2","target":"graph","created_at":"2026-07-05T10:20:17Z","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/2411.16502/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reward models (RMs) are a crucial component in the alignment of large language models' (LLMs) outputs with human values. RMs approximate human preferences over possible LLM responses to the same prompt by predicting and comparing reward scores. However, as they are typically modified versions of LLMs with scalar output heads, RMs are large black boxes whose predictions are not explainable. More transparent RMs would enable improved trust in the alignment of LLMs. In this work, we propose to use contrastive explanations to explain any binary response comparison made by an RM. Specifically, we g","authors_text":"Freddy Lecue, Junqi Jiang, Manuela Veloso, Saumitra Mishra, Tom Bewley","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-25T15:37:27Z","title":"Interpreting Language Reward Models via Contrastive Explanations"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.16502","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:dc9502ba3645b936c709ed18d3a9f979a0be10aacf2e8ca3a34382df477ebabb","target":"record","created_at":"2026-07-05T10:20:17Z","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":"613b2d784c3df2681a483e836939e98e171c891045bb82822ee2ce38208f6818","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-25T15:37:27Z","title_canon_sha256":"18c575c81f91d3b35f17629e4a976fcf48b09af6e1836ed8739637510a964047"},"schema_version":"1.0","source":{"id":"2411.16502","kind":"arxiv","version":2}},"canonical_sha256":"12f09c78e94c45e1ce3332518416bf0e9d407c6a3db5ac98743d5aecea5a6a57","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"12f09c78e94c45e1ce3332518416bf0e9d407c6a3db5ac98743d5aecea5a6a57","first_computed_at":"2026-07-05T10:20:17.749282Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:20:17.749282Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"qnYU27NgYoDVqGkrLOe6nUNVx8QMeknHtA9qCPrWryG0hvTTfRQe/KrstSCceONiVY0BAwDXMnWI8M9YhcoKCw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:20:17.749871Z","signed_message":"canonical_sha256_bytes"},"source_id":"2411.16502","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:dc9502ba3645b936c709ed18d3a9f979a0be10aacf2e8ca3a34382df477ebabb","sha256:f1fea8d59b0bed316b048abad6ced4c6d275759c10748857b13a8bab9eb5d0d2"],"state_sha256":"d4903e9bd1c842faddc65625247bf6afd9b5b14cbb5b761b0d4567fee1f9379d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0aoYqtZu6Rf4HXOFlSbVWMhEhSStZXGRXBgPi6EH8TE2A2mUipoAo//wWRd4y1I2EE5naBOMqy7F0CT36ZuXAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T17:20:15.376455Z","bundle_sha256":"d6a9516e691564e3350bf945475a969b2e1846679c454fceb0e28ef3dadd4a19"}}