{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:ZLQBVEFR4L5UZF77PLZSQCPPFT","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":"97cda43c0a1220d30ec59a9ce29938dfb85f6349abd8091576c91419c7f8691b","cross_cats_sorted":["math.OC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-24T05:44:20Z","title_canon_sha256":"b87713781cf3e80088cb8cd72d21eda8e924e1919d25d54a9855b3976c13d6a3"},"schema_version":"1.0","source":{"id":"2605.24879","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.24879","created_at":"2026-05-26T01:04:03Z"},{"alias_kind":"arxiv_version","alias_value":"2605.24879v1","created_at":"2026-05-26T01:04:03Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.24879","created_at":"2026-05-26T01:04:03Z"},{"alias_kind":"pith_short_12","alias_value":"ZLQBVEFR4L5U","created_at":"2026-05-26T01:04:03Z"},{"alias_kind":"pith_short_16","alias_value":"ZLQBVEFR4L5UZF77","created_at":"2026-05-26T01:04:03Z"},{"alias_kind":"pith_short_8","alias_value":"ZLQBVEFR","created_at":"2026-05-26T01:04:03Z"}],"graph_snapshots":[{"event_id":"sha256:9b29f2c8545030c858b67fa5ad7542f42910142aa1ef8bd39631344c61bbae9c","target":"graph","created_at":"2026-05-26T01:04:03Z","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/2605.24879/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are trained on vast datasets that may contain sensitive information. Differential privacy (DP), the de facto standard for formal privacy guarantees, provides a principled framework for training LLMs with provable privacy protection. However, state-of-the-art DP training implementations rely on fast gradient clipping techniques with memory overhead $O(B \\min\\{T^2, d^2\\})$, where $B$ is the batch size, $T$ is the sequence length, and $d$ is the model width. This becomes prohibitive as both model size and context length grow. We propose DP-SGD-RC, a novel variant of D","authors_text":"Devansh Gupta, Enayat Ullah, Huanyu Zhang, Meisam Razaviyayn, Sai Aparna Aketi","cross_cats":["math.OC"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-24T05:44:20Z","title":"Efficient DP-SGD for LLMs with Randomized Clipping"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.24879","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":"sha256:08355e33d8626edc9dde8b4ef0f395dd46c9a7814a7de00e59f5325d67b1da98","target":"record","created_at":"2026-05-26T01:04:03Z","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":"97cda43c0a1220d30ec59a9ce29938dfb85f6349abd8091576c91419c7f8691b","cross_cats_sorted":["math.OC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-24T05:44:20Z","title_canon_sha256":"b87713781cf3e80088cb8cd72d21eda8e924e1919d25d54a9855b3976c13d6a3"},"schema_version":"1.0","source":{"id":"2605.24879","kind":"arxiv","version":1}},"canonical_sha256":"cae01a90b1e2fb4c97ff7af32809ef2cddd5d00b9a5efed27738091235622f79","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"cae01a90b1e2fb4c97ff7af32809ef2cddd5d00b9a5efed27738091235622f79","first_computed_at":"2026-05-26T01:04:03.162882Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-26T01:04:03.162882Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9KgnkVmMCTIhH6W0Sd4u2qd6ijg3QZ8sLF3u9tnbRrrmw9WwYWRHkmAoQeqsHMCSGjNdjH+lPM6jDcBzmyOXCg==","signature_status":"signed_v1","signed_at":"2026-05-26T01:04:03.163731Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.24879","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:08355e33d8626edc9dde8b4ef0f395dd46c9a7814a7de00e59f5325d67b1da98","sha256:9b29f2c8545030c858b67fa5ad7542f42910142aa1ef8bd39631344c61bbae9c"],"state_sha256":"223bbe0eca03969dc42d3103e3c6fc9d6ffbaea977a75659f1cf21d8f8d4060e"}