{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4DMP4G5LXYS2QGT7CDS3CPWGKK","short_pith_number":"pith:4DMP4G5L","schema_version":"1.0","canonical_sha256":"e0d8fe1babbe25a81a7f10e5b13ec6529395a9f027efeebb765e1f18c0f6577d","source":{"kind":"arxiv","id":"2504.16656","version":4},"attestation_state":"computed","paper":{"title":"Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Jiangbo Pei, Jianhao Zhang, Peiyu Wang, Tianyidan Xie, Weijie Qiu, Wei Shen, Xiaokun Wang, Xuchen Song, Yahui Zhou, Yang Liu, Yichen Wei, Yi Peng, Yunzhuo Hao","submitted_at":"2025-04-23T12:24:10Z","abstract_excerpt":"We present Skywork R1V2, a next-generation multimodal reasoning model and a major leap forward from its predecessor, Skywork R1V. At its core, R1V2 introduces a hybrid reinforcement learning paradigm that jointly leverages the Mixed Preference Optimization (MPO) and the Group Relative Policy Optimization (GRPO), which harmonizes reward-model guidance with rule-based strategies, thereby addressing the long-standing challenge of balancing sophisticated reasoning capabilities with broad generalization. To further enhance training efficiency, we propose the Selective Sample Buffer (SSB) mechanism,"},"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":"2504.16656","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-04-23T12:24:10Z","cross_cats_sorted":[],"title_canon_sha256":"caaa7f70652aa275531c1647e66b80fe6675625294fd9f8be0f83a5b594be527","abstract_canon_sha256":"ff23f30f256bf8f9bfd71fdbba5b93cfdf03e0053a0f91d6c3efbad3a2264baa"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:16:57.664135Z","signature_b64":"hG+QxWVckKAo0i3hOpKBXwWMCIVXHTFZkeEGLeALCnjHzA8UcelZN59cnGCA02pmxKwnuXOwOUflECW+2WPaDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e0d8fe1babbe25a81a7f10e5b13ec6529395a9f027efeebb765e1f18c0f6577d","last_reissued_at":"2026-07-05T11:16:57.663580Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:16:57.663580Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Jiangbo Pei, Jianhao Zhang, Peiyu Wang, Tianyidan Xie, Weijie Qiu, Wei Shen, Xiaokun Wang, Xuchen Song, Yahui Zhou, Yang Liu, Yichen Wei, Yi Peng, Yunzhuo Hao","submitted_at":"2025-04-23T12:24:10Z","abstract_excerpt":"We present Skywork R1V2, a next-generation multimodal reasoning model and a major leap forward from its predecessor, Skywork R1V. At its core, R1V2 introduces a hybrid reinforcement learning paradigm that jointly leverages the Mixed Preference Optimization (MPO) and the Group Relative Policy Optimization (GRPO), which harmonizes reward-model guidance with rule-based strategies, thereby addressing the long-standing challenge of balancing sophisticated reasoning capabilities with broad generalization. To further enhance training efficiency, we propose the Selective Sample Buffer (SSB) mechanism,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.16656","kind":"arxiv","version":4},"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/2504.16656/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":"2504.16656","created_at":"2026-07-05T11:16:57.663648+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.16656v4","created_at":"2026-07-05T11:16:57.663648+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.16656","created_at":"2026-07-05T11:16:57.663648+00:00"},{"alias_kind":"pith_short_12","alias_value":"4DMP4G5LXYS2","created_at":"2026-07-05T11:16:57.663648+00:00"},{"alias_kind":"pith_short_16","alias_value":"4DMP4G5LXYS2QGT7","created_at":"2026-07-05T11:16:57.663648+00:00"},{"alias_kind":"pith_short_8","alias_value":"4DMP4G5L","created_at":"2026-07-05T11:16:57.663648+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25906","citing_title":"OracleAnalyser: Analysing Implicit Semantics of Oracle Bone Scripts through MLLMs with Post-training","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23543","citing_title":"VeriEvol: Scaling Multimodal Mathematical Reasoning via Verifiable Evol-Instruct","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2507.06448","citing_title":"Perception-Aware Policy Optimization for Multimodal Reasoning","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2509.02547","citing_title":"The Landscape of Agentic Reinforcement Learning for LLMs: A Survey","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2511.19972","citing_title":"Boosting Reasoning in Large Multimodal Models via Activation Replay","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07274","citing_title":"Structured Role-Aware Policy Optimization for Multimodal Reasoning","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2508.18265","citing_title":"InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency","ref_index":138,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19902","citing_title":"MMCORE: MultiModal COnnection with Representation Aligned Latent Embeddings","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK","json":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK.json","graph_json":"https://pith.science/api/pith-number/4DMP4G5LXYS2QGT7CDS3CPWGKK/graph.json","events_json":"https://pith.science/api/pith-number/4DMP4G5LXYS2QGT7CDS3CPWGKK/events.json","paper":"https://pith.science/paper/4DMP4G5L"},"agent_actions":{"view_html":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK","download_json":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK.json","view_paper":"https://pith.science/paper/4DMP4G5L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.16656&json=true","fetch_graph":"https://pith.science/api/pith-number/4DMP4G5LXYS2QGT7CDS3CPWGKK/graph.json","fetch_events":"https://pith.science/api/pith-number/4DMP4G5LXYS2QGT7CDS3CPWGKK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK/action/storage_attestation","attest_author":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK/action/author_attestation","sign_citation":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK/action/citation_signature","submit_replication":"https://pith.science/pith/4DMP4G5LXYS2QGT7CDS3CPWGKK/action/replication_record"}},"created_at":"2026-07-05T11:16:57.663648+00:00","updated_at":"2026-07-05T11:16:57.663648+00:00"}