{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:ZBWPNYID7E6LWAX5S2RYW26KK7","short_pith_number":"pith:ZBWPNYID","canonical_record":{"source":{"id":"2508.19567","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-27T04:54:33Z","cross_cats_sorted":[],"title_canon_sha256":"534d7c46108f927a777efdc62aa668072eed5ca904bbbf2b67195b03746079ec","abstract_canon_sha256":"fbbf3f5bf9ef6db2d14bfbd78680f40be772ac2e9fdd52a3cb2a9aebf31825df"},"schema_version":"1.0"},"canonical_sha256":"c86cf6e103f93cbb02fd96a38b6bca57db2cbf616d790a271b969e93f4052fd3","source":{"kind":"arxiv","id":"2508.19567","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.19567","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"arxiv_version","alias_value":"2508.19567v1","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.19567","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"pith_short_12","alias_value":"ZBWPNYID7E6L","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"pith_short_16","alias_value":"ZBWPNYID7E6LWAX5","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"pith_short_8","alias_value":"ZBWPNYID","created_at":"2026-07-05T12:00:14Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:ZBWPNYID7E6LWAX5S2RYW26KK7","target":"record","payload":{"canonical_record":{"source":{"id":"2508.19567","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-27T04:54:33Z","cross_cats_sorted":[],"title_canon_sha256":"534d7c46108f927a777efdc62aa668072eed5ca904bbbf2b67195b03746079ec","abstract_canon_sha256":"fbbf3f5bf9ef6db2d14bfbd78680f40be772ac2e9fdd52a3cb2a9aebf31825df"},"schema_version":"1.0"},"canonical_sha256":"c86cf6e103f93cbb02fd96a38b6bca57db2cbf616d790a271b969e93f4052fd3","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:00:14.211611Z","signature_b64":"jNbOaHhMTBAcHinW0iM3wNOKYJTuQ8kk7lVkkUviHn9nSE3BxzIrXDBNpZddI6PrcZ3XfasYVCxL5xPFs19QDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c86cf6e103f93cbb02fd96a38b6bca57db2cbf616d790a271b969e93f4052fd3","last_reissued_at":"2026-07-05T12:00:14.211088Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:00:14.211088Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2508.19567","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":"2026-07-05T12:00:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8jqb3LVNXjvSL3FkIj9hQWRoExncxJitxG/fEAs8ho47xMdfXqz+iirE+/LIWXzc8vs499zEUOMqn+rNvSnPCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T11:51:21.328052Z"},"content_sha256":"6388cd2131d513fbfeedfecdb7fba83784856530e1f76deb2dcd79e8e2e730ea","schema_version":"1.0","event_id":"sha256:6388cd2131d513fbfeedfecdb7fba83784856530e1f76deb2dcd79e8e2e730ea"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:ZBWPNYID7E6LWAX5S2RYW26KK7","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Counterfactual Reward Model Training for Bias Mitigation in Multimodal Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"N Harshit, Sheryl Mathew","submitted_at":"2025-08-27T04:54:33Z","abstract_excerpt":"In reinforcement learning with human feedback (RLHF), reward models can efficiently learn and amplify latent biases within multimodal datasets, which can lead to imperfect policy optimization through flawed reward signals and decreased fairness. Bias mitigation studies have often applied passive constraints, which can fail under causal confounding. Here, we present a counterfactual reward model that introduces causal inference with multimodal representation learning to provide an unsupervised, bias-resilient reward signal. The heart of our contribution is the Counterfactual Trust Score, an agg"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.19567","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/2508.19567/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":"2026-07-05T12:00:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GgRbtB3ulb+PVvQ9qVG4yxh6E4anLgbDiePqH4/Ycn6LI9YxxFKzuoLi8PeTuvgO4O9KcLG/PWn8gghpTZ0TBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T11:51:21.328701Z"},"content_sha256":"1cde77ebdbd7b3f6a580f217c0fc35e5a8d39daa81c36939de4898eeb7f47aa9","schema_version":"1.0","event_id":"sha256:1cde77ebdbd7b3f6a580f217c0fc35e5a8d39daa81c36939de4898eeb7f47aa9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ZBWPNYID7E6LWAX5S2RYW26KK7/bundle.json","state_url":"https://pith.science/pith/ZBWPNYID7E6LWAX5S2RYW26KK7/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ZBWPNYID7E6LWAX5S2RYW26KK7/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":"2026-08-06T11:51:21Z","links":{"resolver":"https://pith.science/pith/ZBWPNYID7E6LWAX5S2RYW26KK7","bundle":"https://pith.science/pith/ZBWPNYID7E6LWAX5S2RYW26KK7/bundle.json","state":"https://pith.science/pith/ZBWPNYID7E6LWAX5S2RYW26KK7/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ZBWPNYID7E6LWAX5S2RYW26KK7/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:ZBWPNYID7E6LWAX5S2RYW26KK7","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":"fbbf3f5bf9ef6db2d14bfbd78680f40be772ac2e9fdd52a3cb2a9aebf31825df","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-27T04:54:33Z","title_canon_sha256":"534d7c46108f927a777efdc62aa668072eed5ca904bbbf2b67195b03746079ec"},"schema_version":"1.0","source":{"id":"2508.19567","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.19567","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"arxiv_version","alias_value":"2508.19567v1","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.19567","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"pith_short_12","alias_value":"ZBWPNYID7E6L","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"pith_short_16","alias_value":"ZBWPNYID7E6LWAX5","created_at":"2026-07-05T12:00:14Z"},{"alias_kind":"pith_short_8","alias_value":"ZBWPNYID","created_at":"2026-07-05T12:00:14Z"}],"graph_snapshots":[{"event_id":"sha256:1cde77ebdbd7b3f6a580f217c0fc35e5a8d39daa81c36939de4898eeb7f47aa9","target":"graph","created_at":"2026-07-05T12:00:14Z","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/2508.19567/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In reinforcement learning with human feedback (RLHF), reward models can efficiently learn and amplify latent biases within multimodal datasets, which can lead to imperfect policy optimization through flawed reward signals and decreased fairness. Bias mitigation studies have often applied passive constraints, which can fail under causal confounding. Here, we present a counterfactual reward model that introduces causal inference with multimodal representation learning to provide an unsupervised, bias-resilient reward signal. The heart of our contribution is the Counterfactual Trust Score, an agg","authors_text":"N Harshit, Sheryl Mathew","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-27T04:54:33Z","title":"Counterfactual Reward Model Training for Bias Mitigation in Multimodal Reinforcement Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.19567","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":"sha256:6388cd2131d513fbfeedfecdb7fba83784856530e1f76deb2dcd79e8e2e730ea","target":"record","created_at":"2026-07-05T12:00:14Z","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":"fbbf3f5bf9ef6db2d14bfbd78680f40be772ac2e9fdd52a3cb2a9aebf31825df","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-27T04:54:33Z","title_canon_sha256":"534d7c46108f927a777efdc62aa668072eed5ca904bbbf2b67195b03746079ec"},"schema_version":"1.0","source":{"id":"2508.19567","kind":"arxiv","version":1}},"canonical_sha256":"c86cf6e103f93cbb02fd96a38b6bca57db2cbf616d790a271b969e93f4052fd3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c86cf6e103f93cbb02fd96a38b6bca57db2cbf616d790a271b969e93f4052fd3","first_computed_at":"2026-07-05T12:00:14.211088Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T12:00:14.211088Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"jNbOaHhMTBAcHinW0iM3wNOKYJTuQ8kk7lVkkUviHn9nSE3BxzIrXDBNpZddI6PrcZ3XfasYVCxL5xPFs19QDA==","signature_status":"signed_v1","signed_at":"2026-07-05T12:00:14.211611Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.19567","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6388cd2131d513fbfeedfecdb7fba83784856530e1f76deb2dcd79e8e2e730ea","sha256:1cde77ebdbd7b3f6a580f217c0fc35e5a8d39daa81c36939de4898eeb7f47aa9"],"state_sha256":"f318c51e2dba6e61bf3f08d3f3a4da447117e90b68bec03e783543a4cae223ce"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ECCmIdw8MOMJRzBLjttUUZhwKtIX+KvJH1Ne0eKCBihvat/omNsZYalKf9RhhyA2yiNznZnmiKh0IJV2bWwHCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T11:51:21.333586Z","bundle_sha256":"2e114af965e7c1440831ada854d63cba0843d32e3d49d1d547b7234c02d687d5"}}