{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KALDCBDNNOWWQKIFZ6QLKDARYW","short_pith_number":"pith:KALDCBDN","schema_version":"1.0","canonical_sha256":"501631046d6bad682905cfa0b50c11c58e1f03890ce95375f2f6427d71a887f6","source":{"kind":"arxiv","id":"2601.05877","version":3},"attestation_state":"computed","paper":{"title":"iReasoner: Trajectory-Aware Intrinsic Reasoning Supervision for Self-Evolving Large Multimodal Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Manikandarajan Venmathimaran, Meghana Sunil, Muthu Subash Kavitha","submitted_at":"2026-01-09T15:53:42Z","abstract_excerpt":"Recent work shows that large multimodal models (LMMs) can self-improve from unlabeled data via self-play and intrinsic feedback. Yet existing self-evolving frameworks mainly reward final outcomes, leaving intermediate reasoning weakly constrained despite its importance for visually grounded decision making. We propose iReasoner, a self-evolving framework that improves an LMM's implicit reasoning by explicitly eliciting chain-of-thought (CoT) and rewarding its internal agreement. In a Proposer--Solver loop over unlabeled images, iReasoner augments outcome-level intrinsic rewards with a trajecto"},"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":"2601.05877","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-01-09T15:53:42Z","cross_cats_sorted":[],"title_canon_sha256":"90fffceffa4c2b839cfadc9a6b1464c9ea5fb7a61e48a50b55dd0bd1d226d92b","abstract_canon_sha256":"d2d19e0cfad04bf93083b02bbf35b3dd7b7a3a8ee519c522b805f8829da13682"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:04:20.764899Z","signature_b64":"LjdpU23Ki4bZo3J+OyDRilT0kfhLfRufFnXx45JvbGyQ3Ux815RCE2/Xhy/rYr13VbZorx//DVSWlR4q8OduBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"501631046d6bad682905cfa0b50c11c58e1f03890ce95375f2f6427d71a887f6","last_reissued_at":"2026-05-21T01:04:20.763994Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:04:20.763994Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"iReasoner: Trajectory-Aware Intrinsic Reasoning Supervision for Self-Evolving Large Multimodal Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Manikandarajan Venmathimaran, Meghana Sunil, Muthu Subash Kavitha","submitted_at":"2026-01-09T15:53:42Z","abstract_excerpt":"Recent work shows that large multimodal models (LMMs) can self-improve from unlabeled data via self-play and intrinsic feedback. Yet existing self-evolving frameworks mainly reward final outcomes, leaving intermediate reasoning weakly constrained despite its importance for visually grounded decision making. We propose iReasoner, a self-evolving framework that improves an LMM's implicit reasoning by explicitly eliciting chain-of-thought (CoT) and rewarding its internal agreement. In a Proposer--Solver loop over unlabeled images, iReasoner augments outcome-level intrinsic rewards with a trajecto"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.05877","kind":"arxiv","version":3},"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/2601.05877/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":"2601.05877","created_at":"2026-05-21T01:04:20.764129+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.05877v3","created_at":"2026-05-21T01:04:20.764129+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.05877","created_at":"2026-05-21T01:04:20.764129+00:00"},{"alias_kind":"pith_short_12","alias_value":"KALDCBDNNOWW","created_at":"2026-05-21T01:04:20.764129+00:00"},{"alias_kind":"pith_short_16","alias_value":"KALDCBDNNOWWQKIF","created_at":"2026-05-21T01:04:20.764129+00:00"},{"alias_kind":"pith_short_8","alias_value":"KALDCBDN","created_at":"2026-05-21T01:04:20.764129+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.21931","citing_title":"EvoVid: Temporal-Centric Self-Evolution for Video Large Language Models","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18320","citing_title":"EVE: Verifiable Self-Evolution of MLLMs via Executable Visual Transformations","ref_index":39,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW","json":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW.json","graph_json":"https://pith.science/api/pith-number/KALDCBDNNOWWQKIFZ6QLKDARYW/graph.json","events_json":"https://pith.science/api/pith-number/KALDCBDNNOWWQKIFZ6QLKDARYW/events.json","paper":"https://pith.science/paper/KALDCBDN"},"agent_actions":{"view_html":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW","download_json":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW.json","view_paper":"https://pith.science/paper/KALDCBDN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.05877&json=true","fetch_graph":"https://pith.science/api/pith-number/KALDCBDNNOWWQKIFZ6QLKDARYW/graph.json","fetch_events":"https://pith.science/api/pith-number/KALDCBDNNOWWQKIFZ6QLKDARYW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW/action/storage_attestation","attest_author":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW/action/author_attestation","sign_citation":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW/action/citation_signature","submit_replication":"https://pith.science/pith/KALDCBDNNOWWQKIFZ6QLKDARYW/action/replication_record"}},"created_at":"2026-05-21T01:04:20.764129+00:00","updated_at":"2026-05-21T01:04:20.764129+00:00"}