{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:O644HULKCS64KLV6DTJKJNVMVI","short_pith_number":"pith:O644HULK","schema_version":"1.0","canonical_sha256":"77b9c3d16a14bdc52ebe1cd2a4b6acaa131fe678f41b06380a024ff695e101ad","source":{"kind":"arxiv","id":"2512.15702","version":2},"attestation_state":"computed","paper":{"title":"End-to-End Training for Autoregressive Video Diffusion via Self-Resampling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Ceyuan Yang, Dahua Lin, Hao He, Meng Wei, Weilin Huang, Yang Zhao, Yuwei Guo, Zhenheng Yang","submitted_at":"2025-12-17T18:53:29Z","abstract_excerpt":"Autoregressive video diffusion models hold promise for world simulation but are vulnerable to exposure bias arising from the train-test mismatch. While recent works address this via post-training, they typically rely on a bidirectional teacher model or discriminator. To achieve an end-to-end solution, we introduce Resampling Forcing, a teacher-free framework that enables training autoregressive video models from scratch and at scale. Central to our approach is a self-resampling scheme that simulates inference-time model errors on history frames during training. Conditioned on these degraded hi"},"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":"2512.15702","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-12-17T18:53:29Z","cross_cats_sorted":[],"title_canon_sha256":"ee458844bad601f8c1efdc721825e6a6d227d780afc4e34c5e06ee15eb388093","abstract_canon_sha256":"2c6ba361bf1c7ba53352c564307647a5887c4e441b09c0b596c3efd682e8d013"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-02T01:17:26.415064Z","signature_b64":"CALYyara9/iRY9bByNlldoPK/ZqD1dG7P9l13tuy+0HINUcxcJmq/KQYYnVEFeCJwSTSTjDp6N8Wuc00uggkCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"77b9c3d16a14bdc52ebe1cd2a4b6acaa131fe678f41b06380a024ff695e101ad","last_reissued_at":"2026-07-02T01:17:26.414505Z","signature_status":"signed_v1","first_computed_at":"2026-07-02T01:17:26.414505Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"End-to-End Training for Autoregressive Video Diffusion via Self-Resampling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Ceyuan Yang, Dahua Lin, Hao He, Meng Wei, Weilin Huang, Yang Zhao, Yuwei Guo, Zhenheng Yang","submitted_at":"2025-12-17T18:53:29Z","abstract_excerpt":"Autoregressive video diffusion models hold promise for world simulation but are vulnerable to exposure bias arising from the train-test mismatch. While recent works address this via post-training, they typically rely on a bidirectional teacher model or discriminator. To achieve an end-to-end solution, we introduce Resampling Forcing, a teacher-free framework that enables training autoregressive video models from scratch and at scale. Central to our approach is a self-resampling scheme that simulates inference-time model errors on history frames during training. Conditioned on these degraded hi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.15702","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/2512.15702/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":"2512.15702","created_at":"2026-07-02T01:17:26.414568+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.15702v2","created_at":"2026-07-02T01:17:26.414568+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.15702","created_at":"2026-07-02T01:17:26.414568+00:00"},{"alias_kind":"pith_short_12","alias_value":"O644HULKCS64","created_at":"2026-07-02T01:17:26.414568+00:00"},{"alias_kind":"pith_short_16","alias_value":"O644HULKCS64KLV6","created_at":"2026-07-02T01:17:26.414568+00:00"},{"alias_kind":"pith_short_8","alias_value":"O644HULK","created_at":"2026-07-02T01:17:26.414568+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":17,"internal_anchor_count":17,"sample":[{"citing_arxiv_id":"2607.01060","citing_title":"RoboWorld: Fast and Reliable Neural Simulators for Generalist Robot Policy Evaluation","ref_index":66,"is_internal_anchor":true},{"citing_arxiv_id":"2605.30855","citing_title":"Robust Dreamer: Deviation-Aware Latent Gaussian Memory for Action-Controlled AR Video Generation","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15190","citing_title":"RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPO","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15141","citing_title":"Causal Forcing++: Scalable Few-Step Autoregressive Diffusion Distillation for Real-Time Interactive Video Generation","ref_index":46,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21028","citing_title":"DySink: Dynamic Frame Sinks for Autoregressive Long Video Generation","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11596","citing_title":"HorizonDrive: Self-Corrective Autoregressive World Model for Long-horizon Driving Simulation","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2602.02214","citing_title":"Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22144","citing_title":"One Sentence, One Drama: Personalized Short-Form Drama Generation via Multi-Agent Systems","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2602.02214","citing_title":"Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21028","citing_title":"DySink: Dynamic Frame Sinks for Autoregressive Long Video Generation","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2602.07775","citing_title":"Rolling Sink: Bridging Limited-Horizon Training and Open-Ended Testing in Autoregressive Video Diffusion","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14487","citing_title":"Head Forcing: Long Autoregressive Video Generation via Head Heterogeneity","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13724","citing_title":"AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12496","citing_title":"CausalCine: Real-Time Autoregressive Generation for Multi-Shot Video Narratives","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11596","citing_title":"HorizonDrive: Self-Corrective Autoregressive World Model for Long-horizon Driving Simulation","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06939","citing_title":"Grounded Forcing: Bridging Time-Independent Semantics and Proximal Dynamics in Autoregressive Video Synthesis","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2604.15911","citing_title":"Efficient Video Diffusion Models: Advancements and Challenges","ref_index":283,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI","json":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI.json","graph_json":"https://pith.science/api/pith-number/O644HULKCS64KLV6DTJKJNVMVI/graph.json","events_json":"https://pith.science/api/pith-number/O644HULKCS64KLV6DTJKJNVMVI/events.json","paper":"https://pith.science/paper/O644HULK"},"agent_actions":{"view_html":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI","download_json":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI.json","view_paper":"https://pith.science/paper/O644HULK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.15702&json=true","fetch_graph":"https://pith.science/api/pith-number/O644HULKCS64KLV6DTJKJNVMVI/graph.json","fetch_events":"https://pith.science/api/pith-number/O644HULKCS64KLV6DTJKJNVMVI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI/action/storage_attestation","attest_author":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI/action/author_attestation","sign_citation":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI/action/citation_signature","submit_replication":"https://pith.science/pith/O644HULKCS64KLV6DTJKJNVMVI/action/replication_record"}},"created_at":"2026-07-02T01:17:26.414568+00:00","updated_at":"2026-07-02T01:17:26.414568+00:00"}