{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:3R3RAISTDKFJS36FY7L4YL3IVJ","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":"0401f827309c542b473c6e5746e57e61bcd0b5a29b266c6b0594255d825152f4","cross_cats_sorted":["math.OC","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-30T05:11:13Z","title_canon_sha256":"8004a7a4102d496e89fd8b3701d515120121a871ce920e3947ba9f312f55beac"},"schema_version":"1.0","source":{"id":"2606.00539","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.00539","created_at":"2026-06-02T01:03:57Z"},{"alias_kind":"arxiv_version","alias_value":"2606.00539v1","created_at":"2026-06-02T01:03:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.00539","created_at":"2026-06-02T01:03:57Z"},{"alias_kind":"pith_short_12","alias_value":"3R3RAISTDKFJ","created_at":"2026-06-02T01:03:57Z"},{"alias_kind":"pith_short_16","alias_value":"3R3RAISTDKFJS36F","created_at":"2026-06-02T01:03:57Z"},{"alias_kind":"pith_short_8","alias_value":"3R3RAIST","created_at":"2026-06-02T01:03:57Z"}],"graph_snapshots":[{"event_id":"sha256:ab422df4b8bb6ed342aac35628793ff302841327615b426315d83ea730d60f68","target":"graph","created_at":"2026-06-02T01:03:57Z","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/2606.00539/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Training stability is a key bottleneck in low-precision language model training: efficient low-cost paths can still produce short-lived numerical risks at a small set of operators. We formulate this as runtime stability control and present Gradient Norm-to-Mean Ratio (GNMR), a lightweight controller that compares each recoverable unit's current gradient norm with its historical mean. Together with $\\Delta$-GNMR for abrupt short-window increases, GNMR maps local risk signals to bounded recovery actions under a hard $\\mathrm{maxO}$ budget and a short lock interval, without changing the numerical","authors_text":"Boao Kong, Engao Zhang, Guohong Li, Kun Yuan, Weichen Jia, Yao Wang, Yaoyuan Wang, Yonghan Dong, Yunke Peng","cross_cats":["math.OC","stat.ML"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-30T05:11:13Z","title":"GNMR: Runtime Stability Control for Low-Precision Large Language Model Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.00539","kind":"arxiv","version":1},"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":"sha256:336747a13f1a65e66071bdace0e51ae450423e34d32efcb69c59bb1f2c6ea539","target":"record","created_at":"2026-06-02T01:03:57Z","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":"0401f827309c542b473c6e5746e57e61bcd0b5a29b266c6b0594255d825152f4","cross_cats_sorted":["math.OC","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-30T05:11:13Z","title_canon_sha256":"8004a7a4102d496e89fd8b3701d515120121a871ce920e3947ba9f312f55beac"},"schema_version":"1.0","source":{"id":"2606.00539","kind":"arxiv","version":1}},"canonical_sha256":"dc771022531a8a996fc5c7d7cc2f68aa7767c624f0fc5dafe2389f1550005941","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"dc771022531a8a996fc5c7d7cc2f68aa7767c624f0fc5dafe2389f1550005941","first_computed_at":"2026-06-02T01:03:57.604143Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-02T01:03:57.604143Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ZAgUxAq+lqSMvEhzPHXzvlseIbT/EUxhN4XRjV3C/MKTWh9M4uY8p5iZNvBkg4DNRcCREeOuvCx0ZwDDEeUMDQ==","signature_status":"signed_v1","signed_at":"2026-06-02T01:03:57.604546Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.00539","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:336747a13f1a65e66071bdace0e51ae450423e34d32efcb69c59bb1f2c6ea539","sha256:ab422df4b8bb6ed342aac35628793ff302841327615b426315d83ea730d60f68"],"state_sha256":"4b23bcf80aad6c9f9e157a477fa356e5c1382d5fc182699e8b01906bb738605e"}