{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:JDURUAVDSXWPBLCSDSGB7KZR2R","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":"593bffff0e2365eaac80c1b925b27ac35f20ae5568ddceb7175a0ec86eda6eb5","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-14T02:24:46Z","title_canon_sha256":"b2a8e643e00a2a42fde144366c302edf531898314362e5fc7e91e727fb8a3b93"},"schema_version":"1.0","source":{"id":"2605.14278","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14278","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14278v1","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14278","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"pith_short_12","alias_value":"JDURUAVDSXWP","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"JDURUAVDSXWPBLCS","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"JDURUAVD","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:454093ecca72389d0b9d234869bb2069e3c9ccc55a9fb7d296e37bfe221aef4e","target":"graph","created_at":"2026-05-17T23:39:10Z","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":"KVPO introduces a causal-semantic exploration paradigm that relocates the source of variation from stochastic noise to the historical KV cache, constructing semantically diverse generation branches that remain strictly on the data manifold, and a velocity-field surrogate policy based on Trajectory Velocity Energy that yields a reward-weighted contrastive objective fully consistent with the native ODE formulation."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That stochastically routing historical KV entries produces semantically diverse generation branches that remain strictly on the data manifold without introducing artifacts or deviating from the model's learned distribution."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"KVPO aligns streaming autoregressive video generators with human preferences via ODE-native GRPO, using KV cache for semantic exploration and TVE for velocity-based policy modeling, yielding gains in quality and alignment."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"KVPO aligns streaming autoregressive video generators by routing semantic variations through the KV cache and modeling policies via velocity energy in ODE space."}],"snapshot_sha256":"3854c0bab6a8440eef79302105917c1f916ce464dae5b55e440e45335c16d341"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Aligning streaming autoregressive (AR) video generators with human preferences is challenging. Existing reinforcement learning methods predominantly rely on noise-based exploration and SDE-based surrogate policies that are mismatched to the deterministic ODE dynamics of distilled AR models, and tend to perturb low-level appearance rather than the high-level semantic storyline progression critical for long-horizon coherence. To address these limitations, we present KVPO, an ODE-native online Group Relative Policy Optimization (GRPO) framework for aligning streaming video generators. For diversi","authors_text":"Jun Zhou, Kaixi Cong, Ruicheng Zhang, Shuiyang Mao, Wei Liu, Xiu Li, Zhizhou Zhong, Zunnan Xu","cross_cats":[],"headline":"KVPO aligns streaming autoregressive video generators by routing semantic variations through the KV cache and modeling policies via velocity energy in ODE space.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-14T02:24:46Z","title":"KVPO: ODE-Native GRPO for Autoregressive Video Alignment via KV Semantic Exploration"},"references":{"count":33,"internal_anchors":12,"resolved_work":33,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"arXiv preprint arXiv:2511.16955 (2025) 3","work_id":"20b57bf7-3f3a-4078-a74b-d8065c609465","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"arXiv preprint arXiv:2603.17461 (2026)","work_id":"20db7b14-851b-4a3b-9b8a-53fe5fca0290","year":2026},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"arXiv preprint arXiv:2603.21299 (2026)","work_id":"d62c2c5a-3956-4522-bb31-438b3c66c6c5","year":2026},{"cited_arxiv_id":"2106.09685","doi":"","is_internal_anchor":true,"ref_index":4,"title":"LoRA: Low-Rank Adaptation of Large Language Models","work_id":"0426219a-789e-4964-adc8-a04538510818","year":2021},{"cited_arxiv_id":"2506.08009","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion","work_id":"53e58ef9-7932-4b83-b757-34ac14db3e0f","year":2025}],"snapshot_sha256":"536eea658fb28bb4fd023b78b136fe998ec051f75f2dc0fa7045ea4cff6b43d6"},"source":{"id":"2605.14278","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T02:35:53.482864Z","id":"f7558614-696f-4ed3-87a9-cce9d31f28a9","model_set":{"reader":"grok-4.3"},"one_line_summary":"KVPO aligns streaming autoregressive video generators with human preferences via ODE-native GRPO, using KV cache for semantic exploration and TVE for velocity-based policy modeling, yielding gains in quality and alignment.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"KVPO aligns streaming autoregressive video generators by routing semantic variations through the KV cache and modeling policies via velocity energy in ODE space.","strongest_claim":"KVPO introduces a causal-semantic exploration paradigm that relocates the source of variation from stochastic noise to the historical KV cache, constructing semantically diverse generation branches that remain strictly on the data manifold, and a velocity-field surrogate policy based on Trajectory Velocity Energy that yields a reward-weighted contrastive objective fully consistent with the native ODE formulation.","weakest_assumption":"That stochastically routing historical KV entries produces semantically diverse generation branches that remain strictly on the data manifold without introducing artifacts or deviating from the model's learned distribution."}},"verdict_id":"f7558614-696f-4ed3-87a9-cce9d31f28a9"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:1dc513f64fc77c560cb10d4b80b136939f768ae3fdbd9466996be26400527b51","target":"record","created_at":"2026-05-17T23:39:10Z","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":"593bffff0e2365eaac80c1b925b27ac35f20ae5568ddceb7175a0ec86eda6eb5","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-14T02:24:46Z","title_canon_sha256":"b2a8e643e00a2a42fde144366c302edf531898314362e5fc7e91e727fb8a3b93"},"schema_version":"1.0","source":{"id":"2605.14278","kind":"arxiv","version":1}},"canonical_sha256":"48e91a02a395ecf0ac521c8c1fab31d451214957c417e859b466b8474f3f7860","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"48e91a02a395ecf0ac521c8c1fab31d451214957c417e859b466b8474f3f7860","first_computed_at":"2026-05-17T23:39:10.316265Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:10.316265Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"vMzS3iQFvsTarYte0YWLROtuM1C2uUoI2WDmHqU4+V8n3C7N5JBRxCQZbilVK192kkSBenvQmwYf4JQD04XJCQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:10.316796Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14278","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1dc513f64fc77c560cb10d4b80b136939f768ae3fdbd9466996be26400527b51","sha256:454093ecca72389d0b9d234869bb2069e3c9ccc55a9fb7d296e37bfe221aef4e"],"state_sha256":"9921db37593fbdc1bbf90f6d3adeb5663b26f0bb23c3b5cb81b4c8fad2c381f5"}