{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:EVIGEFXKTZQOFFQLZ6Y5XWO36Z","short_pith_number":"pith:EVIGEFXK","canonical_record":{"source":{"id":"2605.13803","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T17:25:51Z","cross_cats_sorted":[],"title_canon_sha256":"1f3c58016865f2d0427d7bdf33660b4bf2d489be65239d9f31094989d1ebb763","abstract_canon_sha256":"114145067ac93f8f12e153b1a05c767d2e23e58d496978b752faf6d5978ee62d"},"schema_version":"1.0"},"canonical_sha256":"25506216ea9e60e2960bcfb1dbd9dbf6757bdab43fad1292f5e4c1cdde2ad65c","source":{"kind":"arxiv","id":"2605.13803","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13803","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13803v1","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13803","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"pith_short_12","alias_value":"EVIGEFXKTZQO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"EVIGEFXKTZQOFFQL","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"EVIGEFXK","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:EVIGEFXKTZQOFFQLZ6Y5XWO36Z","target":"record","payload":{"canonical_record":{"source":{"id":"2605.13803","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T17:25:51Z","cross_cats_sorted":[],"title_canon_sha256":"1f3c58016865f2d0427d7bdf33660b4bf2d489be65239d9f31094989d1ebb763","abstract_canon_sha256":"114145067ac93f8f12e153b1a05c767d2e23e58d496978b752faf6d5978ee62d"},"schema_version":"1.0"},"canonical_sha256":"25506216ea9e60e2960bcfb1dbd9dbf6757bdab43fad1292f5e4c1cdde2ad65c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:15.488428Z","signature_b64":"88sSFY06jPXwswz1FzBMQUOO+CwFqSZKBidzOsJz0TRgUkekswtiZMz3Mkgi6qhwfuCvLS5ES2N/DWCor8aABw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"25506216ea9e60e2960bcfb1dbd9dbf6757bdab43fad1292f5e4c1cdde2ad65c","last_reissued_at":"2026-05-18T02:44:15.487853Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:15.487853Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.13803","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":"2026-05-18T02:44:15Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LPvq8UsO3ASg9EBTThPiLluPVuahUyRFPHOi794TfE6C4xTRzaMr8ldMca+q8Yub2m+DCYmX1tqrPNpoHdMLDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T16:14:05.652687Z"},"content_sha256":"4575993fe29f3c76af68c6dfa9d194e5b81ba42b2aaf0b9753a0075464e55175","schema_version":"1.0","event_id":"sha256:4575993fe29f3c76af68c6dfa9d194e5b81ba42b2aaf0b9753a0075464e55175"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:EVIGEFXKTZQOFFQLZ6Y5XWO36Z","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"EvoGround: Self-Evolving Video Agents for Video Temporal Grounding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Two self-evolving agents learn video temporal grounding from unlabeled videos alone.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Byoung-Tak Zhang, Lorenzo Torresani, Minjoon Jung","submitted_at":"2026-05-13T17:25:51Z","abstract_excerpt":"Video temporal grounding (VTG) takes an untrimmed video and a natural-language query as input and localizes the temporal moment that best matches the query. Existing methods rely on large, task-specific datasets requiring costly manual annotation. We introduce EvoGround, a framework of two coupled self-evolving agents, a proposer and a solver, that learn temporal grounding from raw videos without any human-labeled data. The proposer generates query--moment pairs from raw videos, while the solver learns to ground them and feeds back signals that improve the proposer in return. Through this self"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Trained on 2.5K unlabeled videos, EvoGround matches or surpasses fully supervised models across multiple VTG benchmarks, while emerging as a state-of-the-art fine-grained video captioner without manual labels.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The mutual reinforcement loop between proposer and solver can bootstrap effective temporal grounding and captioning capabilities starting from raw videos and a shared backbone without any initial human supervision or external reward signals.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A proposer-solver agent pair achieves supervised-level video temporal grounding and fine-grained captioning from 2.5K unlabeled videos via self-reinforcing evolution.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Two self-evolving agents learn video temporal grounding from unlabeled videos alone.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"796395d485ef06834c1c052454b595f7113f6af39759cddc7b27bb5ce0c840c0"},"source":{"id":"2605.13803","kind":"arxiv","version":1},"verdict":{"id":"b39179a1-f7bf-4004-b335-63c64be842b7","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:24:40.631477Z","strongest_claim":"Trained on 2.5K unlabeled videos, EvoGround matches or surpasses fully supervised models across multiple VTG benchmarks, while emerging as a state-of-the-art fine-grained video captioner without manual labels.","one_line_summary":"A proposer-solver agent pair achieves supervised-level video temporal grounding and fine-grained captioning from 2.5K unlabeled videos via self-reinforcing evolution.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The mutual reinforcement loop between proposer and solver can bootstrap effective temporal grounding and captioning capabilities starting from raw videos and a shared backbone without any initial human supervision or external reward signals.","pith_extraction_headline":"Two self-evolving agents learn video temporal grounding from unlabeled videos alone."},"references":{"count":69,"sample":[{"doi":"","year":2022,"title":"Modal-specific pseudo query generation for video corpus moment retrieval,","work_id":"8976c7d1-e459-4a52-a8f2-30bdf3ff6019","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Detecting moments and highlights in videos via natural language queries,","work_id":"28515618-525c-4594-9aa4-d398980a7891","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Can i trust your answer? visually grounded video question answering,","work_id":"534671d4-6ee8-42c5-baeb-15da71cc7c0b","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"TOMATO: Assessing Visual Temporal Reasoning Capabilities in Multimodal Foundation Models","work_id":"7a50d850-6338-479d-ba02-65efc1b03f01","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis","work_id":"77fd5ac9-ae98-4846-9d83-e9c73c8f2a52","ref_index":5,"cited_arxiv_id":"2405.21075","is_internal_anchor":true}],"resolved_work":69,"snapshot_sha256":"beb31c4f201c0196644436bf899bd157d6c3b74f04f0bf44e8437ff822f38214","internal_anchors":20},"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":"b39179a1-f7bf-4004-b335-63c64be842b7"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T02:44:15Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"OYzLQ21OLbuYqtQpotMzAu67yovJpoz7Nz78iKha7jTP2ivQVYQSQOanXAfeC3xY939EyLHchBLzL0crZE9gCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T16:14:05.653175Z"},"content_sha256":"08907c63c9fa7561bf1cfd508ac4fa1c3a95c2f7accdf8d17841d770fe633e9b","schema_version":"1.0","event_id":"sha256:08907c63c9fa7561bf1cfd508ac4fa1c3a95c2f7accdf8d17841d770fe633e9b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/EVIGEFXKTZQOFFQLZ6Y5XWO36Z/bundle.json","state_url":"https://pith.science/pith/EVIGEFXKTZQOFFQLZ6Y5XWO36Z/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/EVIGEFXKTZQOFFQLZ6Y5XWO36Z/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":"2026-06-02T16:14:05Z","links":{"resolver":"https://pith.science/pith/EVIGEFXKTZQOFFQLZ6Y5XWO36Z","bundle":"https://pith.science/pith/EVIGEFXKTZQOFFQLZ6Y5XWO36Z/bundle.json","state":"https://pith.science/pith/EVIGEFXKTZQOFFQLZ6Y5XWO36Z/state.json","well_known_bundle":"https://pith.science/.well-known/pith/EVIGEFXKTZQOFFQLZ6Y5XWO36Z/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:EVIGEFXKTZQOFFQLZ6Y5XWO36Z","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":"114145067ac93f8f12e153b1a05c767d2e23e58d496978b752faf6d5978ee62d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T17:25:51Z","title_canon_sha256":"1f3c58016865f2d0427d7bdf33660b4bf2d489be65239d9f31094989d1ebb763"},"schema_version":"1.0","source":{"id":"2605.13803","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13803","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13803v1","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13803","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"pith_short_12","alias_value":"EVIGEFXKTZQO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"EVIGEFXKTZQOFFQL","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"EVIGEFXK","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:08907c63c9fa7561bf1cfd508ac4fa1c3a95c2f7accdf8d17841d770fe633e9b","target":"graph","created_at":"2026-05-18T02:44:15Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Trained on 2.5K unlabeled videos, EvoGround matches or surpasses fully supervised models across multiple VTG benchmarks, while emerging as a state-of-the-art fine-grained video captioner without manual labels."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The mutual reinforcement loop between proposer and solver can bootstrap effective temporal grounding and captioning capabilities starting from raw videos and a shared backbone without any initial human supervision or external reward signals."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"A proposer-solver agent pair achieves supervised-level video temporal grounding and fine-grained captioning from 2.5K unlabeled videos via self-reinforcing evolution."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Two self-evolving agents learn video temporal grounding from unlabeled videos alone."}],"snapshot_sha256":"796395d485ef06834c1c052454b595f7113f6af39759cddc7b27bb5ce0c840c0"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Video temporal grounding (VTG) takes an untrimmed video and a natural-language query as input and localizes the temporal moment that best matches the query. Existing methods rely on large, task-specific datasets requiring costly manual annotation. We introduce EvoGround, a framework of two coupled self-evolving agents, a proposer and a solver, that learn temporal grounding from raw videos without any human-labeled data. The proposer generates query--moment pairs from raw videos, while the solver learns to ground them and feeds back signals that improve the proposer in return. Through this self","authors_text":"Byoung-Tak Zhang, Lorenzo Torresani, Minjoon Jung","cross_cats":[],"headline":"Two self-evolving agents learn video temporal grounding from unlabeled videos alone.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T17:25:51Z","title":"EvoGround: Self-Evolving Video Agents for Video Temporal Grounding"},"references":{"count":69,"internal_anchors":20,"resolved_work":69,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Modal-specific pseudo query generation for video corpus moment retrieval,","work_id":"8976c7d1-e459-4a52-a8f2-30bdf3ff6019","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Detecting moments and highlights in videos via natural language queries,","work_id":"28515618-525c-4594-9aa4-d398980a7891","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Can i trust your answer? visually grounded video question answering,","work_id":"534671d4-6ee8-42c5-baeb-15da71cc7c0b","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"TOMATO: Assessing Visual Temporal Reasoning Capabilities in Multimodal Foundation Models","work_id":"7a50d850-6338-479d-ba02-65efc1b03f01","year":2024},{"cited_arxiv_id":"2405.21075","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis","work_id":"77fd5ac9-ae98-4846-9d83-e9c73c8f2a52","year":2024}],"snapshot_sha256":"beb31c4f201c0196644436bf899bd157d6c3b74f04f0bf44e8437ff822f38214"},"source":{"id":"2605.13803","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T19:24:40.631477Z","id":"b39179a1-f7bf-4004-b335-63c64be842b7","model_set":{"reader":"grok-4.3"},"one_line_summary":"A proposer-solver agent pair achieves supervised-level video temporal grounding and fine-grained captioning from 2.5K unlabeled videos via self-reinforcing evolution.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Two self-evolving agents learn video temporal grounding from unlabeled videos alone.","strongest_claim":"Trained on 2.5K unlabeled videos, EvoGround matches or surpasses fully supervised models across multiple VTG benchmarks, while emerging as a state-of-the-art fine-grained video captioner without manual labels.","weakest_assumption":"The mutual reinforcement loop between proposer and solver can bootstrap effective temporal grounding and captioning capabilities starting from raw videos and a shared backbone without any initial human supervision or external reward signals."}},"verdict_id":"b39179a1-f7bf-4004-b335-63c64be842b7"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4575993fe29f3c76af68c6dfa9d194e5b81ba42b2aaf0b9753a0075464e55175","target":"record","created_at":"2026-05-18T02:44:15Z","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":"114145067ac93f8f12e153b1a05c767d2e23e58d496978b752faf6d5978ee62d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T17:25:51Z","title_canon_sha256":"1f3c58016865f2d0427d7bdf33660b4bf2d489be65239d9f31094989d1ebb763"},"schema_version":"1.0","source":{"id":"2605.13803","kind":"arxiv","version":1}},"canonical_sha256":"25506216ea9e60e2960bcfb1dbd9dbf6757bdab43fad1292f5e4c1cdde2ad65c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"25506216ea9e60e2960bcfb1dbd9dbf6757bdab43fad1292f5e4c1cdde2ad65c","first_computed_at":"2026-05-18T02:44:15.487853Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T02:44:15.487853Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"88sSFY06jPXwswz1FzBMQUOO+CwFqSZKBidzOsJz0TRgUkekswtiZMz3Mkgi6qhwfuCvLS5ES2N/DWCor8aABw==","signature_status":"signed_v1","signed_at":"2026-05-18T02:44:15.488428Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.13803","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4575993fe29f3c76af68c6dfa9d194e5b81ba42b2aaf0b9753a0075464e55175","sha256:08907c63c9fa7561bf1cfd508ac4fa1c3a95c2f7accdf8d17841d770fe633e9b"],"state_sha256":"d2806c566b27a4bab14c57b926ba8eaf1e62767ec0c6d9b7d83477ae208a709f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LWcWZ7FgaOPe4TNf7VOwb7o+R1TSH+D5+76enP0429ZbBI/GzBiCWK4bsWUjg0gmMxdcUNv3bqPMRzo8UhSeCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-02T16:14:05.655431Z","bundle_sha256":"19b07d31626697c2bfcfe3b7e5cbb845341c66e267ddd1274e8aa6e0e853ee05"}}