{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:H7KFSWOKFYLLAY2NQ5UGIQX5IX","short_pith_number":"pith:H7KFSWOK","schema_version":"1.0","canonical_sha256":"3fd45959ca2e16b0634d87686442fd45cf9170f7a243e5487b32aa18dc5e75d3","source":{"kind":"arxiv","id":"2608.08315","version":1},"attestation_state":"computed","paper":{"title":"Your VLM Already Knows When: Training-Free Temporal Grounding by Asking Yes or No","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Barry Devereux, Hui Wang, Ji Huang","submitted_at":"2026-08-08T19:57:47Z","abstract_excerpt":"Multimodal LLMs that recognise events reliably still fail to say when they happen. Prompted for timestamps, strong VLMs reach as little as $3.8\\%$ R@0.5 on Charades-STA, and $77$ to $80\\%$ of their wrong predictions carry low output entropy: the models are confidently wrong, and entropy-based error detection stays below a random classifier. We show that this failure lives in the task interface, not in perception. Holding the weights fixed, replacing timestamp regression with a coarse-to-fine scan of binary questions, whose first-token probabilities are consumed only as a ranking, raises R@0.5 "},"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":"2608.08315","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-08-08T19:57:47Z","cross_cats_sorted":["cs.MM"],"title_canon_sha256":"2ea1fedf7afe0e410a07541f9331b5df914406321eefaf2d4f67becc1b67a477","abstract_canon_sha256":"aa8414921ec6f8fd5402721f9be13142a6f9a618ebed25e576eee6db4eefec97"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T01:22:18.365621Z","signature_b64":"KaBhaEa6xsdo4q+a02S9slHEeYVFyYNj4xh7sB38vPUoo8AeFUAvkUbFT+1YYo87+wdhdeXbap5Nrx0uTTp2BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3fd45959ca2e16b0634d87686442fd45cf9170f7a243e5487b32aa18dc5e75d3","last_reissued_at":"2026-08-11T01:22:18.362646Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T01:22:18.362646Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Your VLM Already Knows When: Training-Free Temporal Grounding by Asking Yes or No","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Barry Devereux, Hui Wang, Ji Huang","submitted_at":"2026-08-08T19:57:47Z","abstract_excerpt":"Multimodal LLMs that recognise events reliably still fail to say when they happen. Prompted for timestamps, strong VLMs reach as little as $3.8\\%$ R@0.5 on Charades-STA, and $77$ to $80\\%$ of their wrong predictions carry low output entropy: the models are confidently wrong, and entropy-based error detection stays below a random classifier. We show that this failure lives in the task interface, not in perception. Holding the weights fixed, replacing timestamp regression with a coarse-to-fine scan of binary questions, whose first-token probabilities are consumed only as a ranking, raises R@0.5 "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.08315","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/2608.08315/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":"2608.08315","created_at":"2026-08-11T01:22:18.363605+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.08315v1","created_at":"2026-08-11T01:22:18.363605+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.08315","created_at":"2026-08-11T01:22:18.363605+00:00"},{"alias_kind":"pith_short_12","alias_value":"H7KFSWOKFYLL","created_at":"2026-08-11T01:22:18.363605+00:00"},{"alias_kind":"pith_short_16","alias_value":"H7KFSWOKFYLLAY2N","created_at":"2026-08-11T01:22:18.363605+00:00"},{"alias_kind":"pith_short_8","alias_value":"H7KFSWOK","created_at":"2026-08-11T01:22:18.363605+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/H7KFSWOKFYLLAY2NQ5UGIQX5IX","json":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX.json","graph_json":"https://pith.science/api/pith-number/H7KFSWOKFYLLAY2NQ5UGIQX5IX/graph.json","events_json":"https://pith.science/api/pith-number/H7KFSWOKFYLLAY2NQ5UGIQX5IX/events.json","paper":"https://pith.science/paper/H7KFSWOK"},"agent_actions":{"view_html":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX","download_json":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX.json","view_paper":"https://pith.science/paper/H7KFSWOK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.08315&json=true","fetch_graph":"https://pith.science/api/pith-number/H7KFSWOKFYLLAY2NQ5UGIQX5IX/graph.json","fetch_events":"https://pith.science/api/pith-number/H7KFSWOKFYLLAY2NQ5UGIQX5IX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX/action/storage_attestation","attest_author":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX/action/author_attestation","sign_citation":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX/action/citation_signature","submit_replication":"https://pith.science/pith/H7KFSWOKFYLLAY2NQ5UGIQX5IX/action/replication_record"}},"created_at":"2026-08-11T01:22:18.363605+00:00","updated_at":"2026-08-11T01:22:18.363605+00:00"}