{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:UZ2J6ZWGSBSBCDLR4TBICFOFJE","short_pith_number":"pith:UZ2J6ZWG","canonical_record":{"source":{"id":"2608.08802","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-09T16:36:42Z","cross_cats_sorted":[],"title_canon_sha256":"a6a66f22b8f1bb55cca390fdd5c521c927f99b36d72459cc841404c413339ceb","abstract_canon_sha256":"852e706141423ff992834211430e23b34545bec454d115374ede8e1774a21f76"},"schema_version":"1.0"},"canonical_sha256":"a6749f66c69064110d71e4c28115c54909306232c7fe5018912257da3fd9d5ce","source":{"kind":"arxiv","id":"2608.08802","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.08802","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"arxiv_version","alias_value":"2608.08802v1","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.08802","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"pith_short_12","alias_value":"UZ2J6ZWGSBSB","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"pith_short_16","alias_value":"UZ2J6ZWGSBSBCDLR","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"pith_short_8","alias_value":"UZ2J6ZWG","created_at":"2026-08-11T01:25:23Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:UZ2J6ZWGSBSBCDLR4TBICFOFJE","target":"record","payload":{"canonical_record":{"source":{"id":"2608.08802","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-09T16:36:42Z","cross_cats_sorted":[],"title_canon_sha256":"a6a66f22b8f1bb55cca390fdd5c521c927f99b36d72459cc841404c413339ceb","abstract_canon_sha256":"852e706141423ff992834211430e23b34545bec454d115374ede8e1774a21f76"},"schema_version":"1.0"},"canonical_sha256":"a6749f66c69064110d71e4c28115c54909306232c7fe5018912257da3fd9d5ce","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T01:25:23.099120Z","signature_b64":"ru10o48sisVxCuqAi0s+W7vU7k3zANyFlkc4SnLWZuTCSlcOWumBhdg0SgtEesFgToib1lIgvBpIGm7JcfHpBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a6749f66c69064110d71e4c28115c54909306232c7fe5018912257da3fd9d5ce","last_reissued_at":"2026-08-11T01:25:23.096423Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T01:25:23.096423Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2608.08802","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":"2026-08-11T01:25:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"kS/iOzrZkEm7NQrATWx8xorvvHQp/fH9kLpZ6Or8XlbjSmlKBszaELA0jPhCMjbC3c37HWMZPnIebrYXmXp1Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T22:38:29.585723Z"},"content_sha256":"a0f2b91cfb6ed36f32afc7db59dc98fd73baf8f8fdc4dedb5c5f0e8bb9ad4d30","schema_version":"1.0","event_id":"sha256:a0f2b91cfb6ed36f32afc7db59dc98fd73baf8f8fdc4dedb5c5f0e8bb9ad4d30"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:UZ2J6ZWGSBSBCDLR4TBICFOFJE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Improving Generalization Robustness of Multimodal RLVR","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Bin Xu, Bohan Zhuang, Chenrui Zhou, Fan Wang, Jiajun Song, Jiasheng Tang, Lama Moukheiber, Pengfei Zhou, Wangbo Zhao, Xiaopeng Peng, Yang You, Yixing Ma, Zhenglin Wan, Zhiwei Tang","submitted_at":"2026-08-09T16:36:42Z","abstract_excerpt":"Reinforcement Learning with Verifiable Rewards (RLVR) makes Multimodal Large Language Models more accurate, but the gains are brittle: simply paraphrasing a question or changing the prompt template can degrade them, which challenges reliable deployment in high-stakes scenarios like medical VQA. We trace this to two issues of the standard RL objective. First, the binary verifier conflates format with content, so the reward signal cannot tell a wrong answer apart from a misformatted one. Second, the training distribution covers only a thin slice of the real-world prompts that the model might mee"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.08802","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/2608.08802/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":"2026-08-11T01:25:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"C/s/E9vUayOj2dBYf3BAsPZEd/75KIqTZs4zJgdwWGMz+3ps/bCPVgc4bpQxSyZBp9o865IvbKto2jHGFuKkDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T22:38:29.586214Z"},"content_sha256":"98021a6ff9afd01b5350faeaa3b310225baa54915b7a71301ccd5fbfe1677acc","schema_version":"1.0","event_id":"sha256:98021a6ff9afd01b5350faeaa3b310225baa54915b7a71301ccd5fbfe1677acc"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/UZ2J6ZWGSBSBCDLR4TBICFOFJE/bundle.json","state_url":"https://pith.science/pith/UZ2J6ZWGSBSBCDLR4TBICFOFJE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/UZ2J6ZWGSBSBCDLR4TBICFOFJE/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":"2026-08-14T22:38:29Z","links":{"resolver":"https://pith.science/pith/UZ2J6ZWGSBSBCDLR4TBICFOFJE","bundle":"https://pith.science/pith/UZ2J6ZWGSBSBCDLR4TBICFOFJE/bundle.json","state":"https://pith.science/pith/UZ2J6ZWGSBSBCDLR4TBICFOFJE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/UZ2J6ZWGSBSBCDLR4TBICFOFJE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:UZ2J6ZWGSBSBCDLR4TBICFOFJE","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":"852e706141423ff992834211430e23b34545bec454d115374ede8e1774a21f76","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-09T16:36:42Z","title_canon_sha256":"a6a66f22b8f1bb55cca390fdd5c521c927f99b36d72459cc841404c413339ceb"},"schema_version":"1.0","source":{"id":"2608.08802","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.08802","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"arxiv_version","alias_value":"2608.08802v1","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.08802","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"pith_short_12","alias_value":"UZ2J6ZWGSBSB","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"pith_short_16","alias_value":"UZ2J6ZWGSBSBCDLR","created_at":"2026-08-11T01:25:23Z"},{"alias_kind":"pith_short_8","alias_value":"UZ2J6ZWG","created_at":"2026-08-11T01:25:23Z"}],"graph_snapshots":[{"event_id":"sha256:98021a6ff9afd01b5350faeaa3b310225baa54915b7a71301ccd5fbfe1677acc","target":"graph","created_at":"2026-08-11T01:25:23Z","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/2608.08802/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement Learning with Verifiable Rewards (RLVR) makes Multimodal Large Language Models more accurate, but the gains are brittle: simply paraphrasing a question or changing the prompt template can degrade them, which challenges reliable deployment in high-stakes scenarios like medical VQA. We trace this to two issues of the standard RL objective. First, the binary verifier conflates format with content, so the reward signal cannot tell a wrong answer apart from a misformatted one. Second, the training distribution covers only a thin slice of the real-world prompts that the model might mee","authors_text":"Bin Xu, Bohan Zhuang, Chenrui Zhou, Fan Wang, Jiajun Song, Jiasheng Tang, Lama Moukheiber, Pengfei Zhou, Wangbo Zhao, Xiaopeng Peng, Yang You, Yixing Ma, Zhenglin Wan, Zhiwei Tang","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-09T16:36:42Z","title":"Improving Generalization Robustness of Multimodal RLVR"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.08802","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:a0f2b91cfb6ed36f32afc7db59dc98fd73baf8f8fdc4dedb5c5f0e8bb9ad4d30","target":"record","created_at":"2026-08-11T01:25:23Z","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":"852e706141423ff992834211430e23b34545bec454d115374ede8e1774a21f76","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-09T16:36:42Z","title_canon_sha256":"a6a66f22b8f1bb55cca390fdd5c521c927f99b36d72459cc841404c413339ceb"},"schema_version":"1.0","source":{"id":"2608.08802","kind":"arxiv","version":1}},"canonical_sha256":"a6749f66c69064110d71e4c28115c54909306232c7fe5018912257da3fd9d5ce","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a6749f66c69064110d71e4c28115c54909306232c7fe5018912257da3fd9d5ce","first_computed_at":"2026-08-11T01:25:23.096423Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-08-11T01:25:23.096423Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ru10o48sisVxCuqAi0s+W7vU7k3zANyFlkc4SnLWZuTCSlcOWumBhdg0SgtEesFgToib1lIgvBpIGm7JcfHpBw==","signature_status":"signed_v1","signed_at":"2026-08-11T01:25:23.099120Z","signed_message":"canonical_sha256_bytes"},"source_id":"2608.08802","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a0f2b91cfb6ed36f32afc7db59dc98fd73baf8f8fdc4dedb5c5f0e8bb9ad4d30","sha256:98021a6ff9afd01b5350faeaa3b310225baa54915b7a71301ccd5fbfe1677acc"],"state_sha256":"bfd432513547f47713ecbe6a80274bcbdd2fb3ac72b894f208180b933a5f22ac"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ppizkw4kQAKbp19ENFvRWYroMr0VCynhFUTCt3yuGWvSmQi1S4mEv9C+2YGtYN3gyalWsU++HU2FOzlTAv/EDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-14T22:38:29.590149Z","bundle_sha256":"07d70fee51d8b04e14398cad435983cda50b32532a1f8f8134800c5dababd133"}}