{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:6FTVDJRX56SNIX5JNVXN3E727K","short_pith_number":"pith:6FTVDJRX","schema_version":"1.0","canonical_sha256":"f16751a637efa4d45fa96d6edd93fafabbe883dfc07ba0be3ad33139f1fc6cfb","source":{"kind":"arxiv","id":"2602.00846","version":2},"attestation_state":"computed","paper":{"title":"Omni-RRM: Advancing Omni Reward Modeling via Automatic Rubric-Grounded Preference Synthesis","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alven Yang, Dehua Ma, Fuqing Bie, Haoran Wang, Huijia Wu, Jian Zhao, Liuyu Xiang, Yiwei Ru, Zhaofeng He, Zhenbo Xu, Zicheng Kong, Zixuan Zhou","submitted_at":"2026-01-31T18:20:45Z","abstract_excerpt":"Multimodal large language models (MLLMs) struggle with alignment due to the limitations of existing reward models (RMs), which are predominantly vision-centric, dependent on costly human labels, and provide opaque scalar scores that fail to capture nuanced reasoning, leading to brittle alignment. We present Omni-RRM, an \\textbf{Omni}-modal \\textbf{R}ubric-grounded \\textbf{R}eward \\textbf{M}odel that generates multi-dimensional reward signals across text, image, video, and audio. To overcome the high cost and inherent inconsistency of human-centric evaluation in multi-dimensional reasoning, we "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2602.00846","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-01-31T18:20:45Z","cross_cats_sorted":[],"title_canon_sha256":"f271e9f3fbef7eb5c04170a08f40ea8f8aac61c494efb45d11330f4481fd4982","abstract_canon_sha256":"e0ddf257ff13e781a2a5a042f8758f449d222a88049c0eabd347c87060837469"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-08T01:19:09.926308Z","signature_b64":"WAlNMaNT46Tj7T93SeE5DzpkLNByTkau86jVjmLFAeWH3camMHpG8eCiliSQsQLNtF52WckQbzSHKSQd4FulCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f16751a637efa4d45fa96d6edd93fafabbe883dfc07ba0be3ad33139f1fc6cfb","last_reissued_at":"2026-07-08T01:19:09.925642Z","signature_status":"signed_v1","first_computed_at":"2026-07-08T01:19:09.925642Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Omni-RRM: Advancing Omni Reward Modeling via Automatic Rubric-Grounded Preference Synthesis","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alven Yang, Dehua Ma, Fuqing Bie, Haoran Wang, Huijia Wu, Jian Zhao, Liuyu Xiang, Yiwei Ru, Zhaofeng He, Zhenbo Xu, Zicheng Kong, Zixuan Zhou","submitted_at":"2026-01-31T18:20:45Z","abstract_excerpt":"Multimodal large language models (MLLMs) struggle with alignment due to the limitations of existing reward models (RMs), which are predominantly vision-centric, dependent on costly human labels, and provide opaque scalar scores that fail to capture nuanced reasoning, leading to brittle alignment. We present Omni-RRM, an \\textbf{Omni}-modal \\textbf{R}ubric-grounded \\textbf{R}eward \\textbf{M}odel that generates multi-dimensional reward signals across text, image, video, and audio. To overcome the high cost and inherent inconsistency of human-centric evaluation in multi-dimensional reasoning, we "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.00846","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/2602.00846/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2602.00846","created_at":"2026-07-08T01:19:09.925713+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.00846v2","created_at":"2026-07-08T01:19:09.925713+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.00846","created_at":"2026-07-08T01:19:09.925713+00:00"},{"alias_kind":"pith_short_12","alias_value":"6FTVDJRX56SN","created_at":"2026-07-08T01:19:09.925713+00:00"},{"alias_kind":"pith_short_16","alias_value":"6FTVDJRX56SNIX5J","created_at":"2026-07-08T01:19:09.925713+00:00"},{"alias_kind":"pith_short_8","alias_value":"6FTVDJRX","created_at":"2026-07-08T01:19:09.925713+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2606.02113","citing_title":"A Primer in Post-Training Reasoning Data: What We Know About How It Works","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2605.30010","citing_title":"EarlyTom: Early Token Compression Completes Fast Video Understanding","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.30244","citing_title":"Reinforcement Learning with Robust Rubric Rewards","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2604.13029","citing_title":"Visual Preference Optimization with Rubric Rewards","ref_index":40,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K","json":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K.json","graph_json":"https://pith.science/api/pith-number/6FTVDJRX56SNIX5JNVXN3E727K/graph.json","events_json":"https://pith.science/api/pith-number/6FTVDJRX56SNIX5JNVXN3E727K/events.json","paper":"https://pith.science/paper/6FTVDJRX"},"agent_actions":{"view_html":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K","download_json":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K.json","view_paper":"https://pith.science/paper/6FTVDJRX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.00846&json=true","fetch_graph":"https://pith.science/api/pith-number/6FTVDJRX56SNIX5JNVXN3E727K/graph.json","fetch_events":"https://pith.science/api/pith-number/6FTVDJRX56SNIX5JNVXN3E727K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K/action/storage_attestation","attest_author":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K/action/author_attestation","sign_citation":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K/action/citation_signature","submit_replication":"https://pith.science/pith/6FTVDJRX56SNIX5JNVXN3E727K/action/replication_record"}},"created_at":"2026-07-08T01:19:09.925713+00:00","updated_at":"2026-07-08T01:19:09.925713+00:00"}