{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HF3Y5QD6I54RC55GIXNHBPLEPO","short_pith_number":"pith:HF3Y5QD6","schema_version":"1.0","canonical_sha256":"39778ec07e47791177a645da70bd647b844cf423373c5c9b34f383a80f901d0b","source":{"kind":"arxiv","id":"2602.13602","version":2},"attestation_state":"computed","paper":{"title":"Towards Sparse Video Understanding and Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Chenwei Xu, Fan Du, Han Liu, Lie Lu, Manling Li, Pranav Maneriker, Shang Wu, Weijian Li, Zhen Ye, Zhuofan Xia, Zihan Wang","submitted_at":"2026-02-14T04:52:11Z","abstract_excerpt":"We present \\revise (\\underline{Re}asoning with \\underline{Vi}deo \\underline{S}parsity), a multi-round agent for video question answering (VQA). Instead of uniformly sampling frames, \\revise selects a small set of informative frames, maintains a summary-as-state across rounds, and stops early when confident. It supports proprietary vision-language models (VLMs) in a ``plug-and-play'' setting and enables reinforcement fine-tuning for open-source models. For fine-tuning, we introduce EAGER (Evidence-Adjusted Gain for Efficient Reasoning), an annotation-free reward with three terms: (1) Confidence"},"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":"2602.13602","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-02-14T04:52:11Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ce804cb1ed29a842426a7e603abf62c0f997b0fa718257e548681b6d145c0838","abstract_canon_sha256":"5556914235743b60206ca8f1e69b10426ff1525d34bc94717658e497cfb38a1c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:43.290182Z","signature_b64":"cf0DuLjPpm9ojZvbKJA7YgDVF3HZ0NQYY28bcnQpJK+rRUAbf5KwU0jMwVSMz/68B8P6bklZiwOG+oqv/h+RAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"39778ec07e47791177a645da70bd647b844cf423373c5c9b34f383a80f901d0b","last_reissued_at":"2026-06-02T01:03:43.289728Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:43.289728Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Sparse Video Understanding and Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Chenwei Xu, Fan Du, Han Liu, Lie Lu, Manling Li, Pranav Maneriker, Shang Wu, Weijian Li, Zhen Ye, Zhuofan Xia, Zihan Wang","submitted_at":"2026-02-14T04:52:11Z","abstract_excerpt":"We present \\revise (\\underline{Re}asoning with \\underline{Vi}deo \\underline{S}parsity), a multi-round agent for video question answering (VQA). Instead of uniformly sampling frames, \\revise selects a small set of informative frames, maintains a summary-as-state across rounds, and stops early when confident. It supports proprietary vision-language models (VLMs) in a ``plug-and-play'' setting and enables reinforcement fine-tuning for open-source models. For fine-tuning, we introduce EAGER (Evidence-Adjusted Gain for Efficient Reasoning), an annotation-free reward with three terms: (1) Confidence"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.13602","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/2602.13602/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":"2602.13602","created_at":"2026-06-02T01:03:43.289802+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.13602v2","created_at":"2026-06-02T01:03:43.289802+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.13602","created_at":"2026-06-02T01:03:43.289802+00:00"},{"alias_kind":"pith_short_12","alias_value":"HF3Y5QD6I54R","created_at":"2026-06-02T01:03:43.289802+00:00"},{"alias_kind":"pith_short_16","alias_value":"HF3Y5QD6I54RC55G","created_at":"2026-06-02T01:03:43.289802+00:00"},{"alias_kind":"pith_short_8","alias_value":"HF3Y5QD6","created_at":"2026-06-02T01:03:43.289802+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/HF3Y5QD6I54RC55GIXNHBPLEPO","json":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO.json","graph_json":"https://pith.science/api/pith-number/HF3Y5QD6I54RC55GIXNHBPLEPO/graph.json","events_json":"https://pith.science/api/pith-number/HF3Y5QD6I54RC55GIXNHBPLEPO/events.json","paper":"https://pith.science/paper/HF3Y5QD6"},"agent_actions":{"view_html":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO","download_json":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO.json","view_paper":"https://pith.science/paper/HF3Y5QD6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.13602&json=true","fetch_graph":"https://pith.science/api/pith-number/HF3Y5QD6I54RC55GIXNHBPLEPO/graph.json","fetch_events":"https://pith.science/api/pith-number/HF3Y5QD6I54RC55GIXNHBPLEPO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO/action/storage_attestation","attest_author":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO/action/author_attestation","sign_citation":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO/action/citation_signature","submit_replication":"https://pith.science/pith/HF3Y5QD6I54RC55GIXNHBPLEPO/action/replication_record"}},"created_at":"2026-06-02T01:03:43.289802+00:00","updated_at":"2026-06-02T01:03:43.289802+00:00"}