{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:5KSMKQEQTFDGRQS2P7X6TCKEBS","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":"ff5e630669c2d118d222ab868ce05f72ff3fe46e8b31d152311778b1803470e5","cross_cats_sorted":["cs.AI","cs.IT","math.IT","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-28T20:41:59Z","title_canon_sha256":"791dd4ade81f9a4101a90af7649755dea2033752524de98d8cf21522024ff2ff"},"schema_version":"1.0","source":{"id":"2512.23075","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.23075","created_at":"2026-06-29T01:14:27Z"},{"alias_kind":"arxiv_version","alias_value":"2512.23075v5","created_at":"2026-06-29T01:14:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.23075","created_at":"2026-06-29T01:14:27Z"},{"alias_kind":"pith_short_12","alias_value":"5KSMKQEQTFDG","created_at":"2026-06-29T01:14:27Z"},{"alias_kind":"pith_short_16","alias_value":"5KSMKQEQTFDGRQS2","created_at":"2026-06-29T01:14:27Z"},{"alias_kind":"pith_short_8","alias_value":"5KSMKQEQ","created_at":"2026-06-29T01:14:27Z"}],"graph_snapshots":[{"event_id":"sha256:9dc9588e7ab211746584f256ca0ba26cf68bc69b9a50df8fec38e46d59f77536","target":"graph","created_at":"2026-06-29T01:14:27Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2512.23075/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Policy gradient methods for Large Language Models optimize a policy $\\pi_\\theta$ via a surrogate objective computed from samples of a rollout policy $\\pi_{\\text{roll}}$. However, modern LLM-RL pipelines suffer from unavoidable implementation divergences -- backend discrepancies, Mixture-of-Experts routing discontinuities, and distributed training staleness -- causing off-policy mismatch ($\\pi_{\\text{roll}} \\neq \\pi_\\theta$) and approximation errors between the surrogate and the true objective. We demonstrate that classical trust region bounds on this error scale as $O(T^2)$ with sequence lengt","authors_text":"Baoxiang Wang, Jiacai Liu, Jiawei Xu, Qian Liu, Yingru Li, Yuxuan Tong, Ziniu Li","cross_cats":["cs.AI","cs.IT","math.IT","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-28T20:41:59Z","title":"Trust Region Masking for Long-Horizon LLM Reinforcement Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.23075","kind":"arxiv","version":5},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ecf5ed97fe585ae94d81e322d8581f7c0fe9f3205337951b35f4be79bb33fa06","target":"record","created_at":"2026-06-29T01:14:27Z","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":"ff5e630669c2d118d222ab868ce05f72ff3fe46e8b31d152311778b1803470e5","cross_cats_sorted":["cs.AI","cs.IT","math.IT","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-28T20:41:59Z","title_canon_sha256":"791dd4ade81f9a4101a90af7649755dea2033752524de98d8cf21522024ff2ff"},"schema_version":"1.0","source":{"id":"2512.23075","kind":"arxiv","version":5}},"canonical_sha256":"eaa4c54090994668c25a7fefe989440c8946c1bfd47a873f3b36276b29db8f4d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"eaa4c54090994668c25a7fefe989440c8946c1bfd47a873f3b36276b29db8f4d","first_computed_at":"2026-06-29T01:14:27.352702Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-29T01:14:27.352702Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"3txP8gnw8kWD5WHf+YR4aJ4kZ3SOIPgq4pid0B8DfhESefTxM93NcLF/pyPZibXCznvA+MWlmJLRsbVl9KMyCg==","signature_status":"signed_v1","signed_at":"2026-06-29T01:14:27.353172Z","signed_message":"canonical_sha256_bytes"},"source_id":"2512.23075","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ecf5ed97fe585ae94d81e322d8581f7c0fe9f3205337951b35f4be79bb33fa06","sha256:9dc9588e7ab211746584f256ca0ba26cf68bc69b9a50df8fec38e46d59f77536"],"state_sha256":"5fce9ba03d8dc70ed0fcf32ae69d8b25e9f544a8337b54b6624f823a6452a431"}