{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LJ5CNM6XR66DPEWK6JA6ISKFXW","short_pith_number":"pith:LJ5CNM6X","schema_version":"1.0","canonical_sha256":"5a7a26b3d78fbc3792caf241e44945bd8a704e9678ee1c4f830b98e6e09e0877","source":{"kind":"arxiv","id":"2606.08708","version":1},"attestation_state":"computed","paper":{"title":"PRPO: Perception-Reinforced Policy Optimization via Token-Level Dynamic Advantage Reshaping","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hangyu Li, Kaitao Jiang, Kangning Niu, Mu Xu, Qiming Li, Ruiyan Gong, Tianlun Li, Xiaolong Cheng","submitted_at":"2026-06-07T16:06:29Z","abstract_excerpt":"Reinforcement Learning with Verifiable Rewards (RLVR) has become an effective paradigm for improving the reasoning capability of Large Vision-Language Models (LVLMs). However, existing RLVR methods primarily rely on trajectory-level outcome rewards, which assign identical learning signals across all generated tokens. This coarse-grained credit assignment is fundamentally mismatched to multimodal reasoning, where only a sparse subset of tokens is causally grounded in visual evidence. Consequently, these pivotal perceptual tokens receive weak supervision and are often overwhelmed by language pri"},"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":"2606.08708","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-06-07T16:06:29Z","cross_cats_sorted":[],"title_canon_sha256":"b9e054fafb142e282f00abccbe4053245d3b0c103fe0481c603d000f67ea917f","abstract_canon_sha256":"5d6f66b121cf943aebe8fa38a0002b8de2411d362734c46e9e52e8f184358b90"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:47.947436Z","signature_b64":"mFedTNqT0VMbRov6BRyyNGnx5AmRapk9vcOK3aAnjGjF57QNr6C3lmiXp5jjw7eLoyPWY31/hICB7tsc+qqgBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5a7a26b3d78fbc3792caf241e44945bd8a704e9678ee1c4f830b98e6e09e0877","last_reissued_at":"2026-06-09T01:05:47.947029Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:47.947029Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PRPO: Perception-Reinforced Policy Optimization via Token-Level Dynamic Advantage Reshaping","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hangyu Li, Kaitao Jiang, Kangning Niu, Mu Xu, Qiming Li, Ruiyan Gong, Tianlun Li, Xiaolong Cheng","submitted_at":"2026-06-07T16:06:29Z","abstract_excerpt":"Reinforcement Learning with Verifiable Rewards (RLVR) has become an effective paradigm for improving the reasoning capability of Large Vision-Language Models (LVLMs). However, existing RLVR methods primarily rely on trajectory-level outcome rewards, which assign identical learning signals across all generated tokens. This coarse-grained credit assignment is fundamentally mismatched to multimodal reasoning, where only a sparse subset of tokens is causally grounded in visual evidence. Consequently, these pivotal perceptual tokens receive weak supervision and are often overwhelmed by language pri"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08708","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/2606.08708/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":"2606.08708","created_at":"2026-06-09T01:05:47.947084+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.08708v1","created_at":"2026-06-09T01:05:47.947084+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08708","created_at":"2026-06-09T01:05:47.947084+00:00"},{"alias_kind":"pith_short_12","alias_value":"LJ5CNM6XR66D","created_at":"2026-06-09T01:05:47.947084+00:00"},{"alias_kind":"pith_short_16","alias_value":"LJ5CNM6XR66DPEWK","created_at":"2026-06-09T01:05:47.947084+00:00"},{"alias_kind":"pith_short_8","alias_value":"LJ5CNM6X","created_at":"2026-06-09T01:05:47.947084+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW","json":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW.json","graph_json":"https://pith.science/api/pith-number/LJ5CNM6XR66DPEWK6JA6ISKFXW/graph.json","events_json":"https://pith.science/api/pith-number/LJ5CNM6XR66DPEWK6JA6ISKFXW/events.json","paper":"https://pith.science/paper/LJ5CNM6X"},"agent_actions":{"view_html":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW","download_json":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW.json","view_paper":"https://pith.science/paper/LJ5CNM6X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.08708&json=true","fetch_graph":"https://pith.science/api/pith-number/LJ5CNM6XR66DPEWK6JA6ISKFXW/graph.json","fetch_events":"https://pith.science/api/pith-number/LJ5CNM6XR66DPEWK6JA6ISKFXW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW/action/storage_attestation","attest_author":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW/action/author_attestation","sign_citation":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW/action/citation_signature","submit_replication":"https://pith.science/pith/LJ5CNM6XR66DPEWK6JA6ISKFXW/action/replication_record"}},"created_at":"2026-06-09T01:05:47.947084+00:00","updated_at":"2026-06-09T01:05:47.947084+00:00"}