{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:TKXRKTDGZR47ZXPHD3W6LQS6BA","short_pith_number":"pith:TKXRKTDG","canonical_record":{"source":{"id":"2502.17607","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T19:49:15Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"295bb1c7fd395e0560eea7246249456725b1fae5bb15b3bcf1a2b4ffc6c78151","abstract_canon_sha256":"69f919e61d5758f0313b1569b779a2a294a215052094eda0262119001c4fb2c3"},"schema_version":"1.0"},"canonical_sha256":"9aaf154c66cc79fcdde71eede5c25e0816bc694990babb0b40ab124c16448834","source":{"kind":"arxiv","id":"2502.17607","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.17607","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"arxiv_version","alias_value":"2502.17607v2","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.17607","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"pith_short_12","alias_value":"TKXRKTDGZR47","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"pith_short_16","alias_value":"TKXRKTDGZR47ZXPH","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"pith_short_8","alias_value":"TKXRKTDG","created_at":"2026-07-05T11:17:41Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:TKXRKTDGZR47ZXPHD3W6LQS6BA","target":"record","payload":{"canonical_record":{"source":{"id":"2502.17607","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T19:49:15Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"295bb1c7fd395e0560eea7246249456725b1fae5bb15b3bcf1a2b4ffc6c78151","abstract_canon_sha256":"69f919e61d5758f0313b1569b779a2a294a215052094eda0262119001c4fb2c3"},"schema_version":"1.0"},"canonical_sha256":"9aaf154c66cc79fcdde71eede5c25e0816bc694990babb0b40ab124c16448834","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:41.928156Z","signature_b64":"VAypGMiD842rZ6NOULaZ/ZTxcqgrQD8SpqwNA7VRuU+0xKl9xOrJkpBOMNVY78yVQAUKVgjFfl2/+TfmKx2TDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9aaf154c66cc79fcdde71eede5c25e0816bc694990babb0b40ab124c16448834","last_reissued_at":"2026-07-05T11:17:41.927634Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:41.927634Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.17607","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:17:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DuxIEtV0btjnexvCWLWcvz3WsBKUC8YwXtYiFcP9GQBnje4OwQ0yCPO4lmh7JqG2NzXH3VZbyM2dselVJbH8CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T08:21:25.113729Z"},"content_sha256":"261942f2c4e67dcbe5c03cb3933c2c204ce8173c7bdd4eedd2ab4164a9000655","schema_version":"1.0","event_id":"sha256:261942f2c4e67dcbe5c03cb3933c2c204ce8173c7bdd4eedd2ab4164a9000655"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:TKXRKTDGZR47ZXPHD3W6LQS6BA","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Synthetic Text Generation for Training Large Language Models via Gradient Matching","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Baharan Mirzasoleiman, Dang Nguyen, Meisam Razaviyayn, MohammadHossein Bateni, Vahab Mirrokni, Zeman Li","submitted_at":"2025-02-24T19:49:15Z","abstract_excerpt":"Synthetic data has the potential to improve the performance, training efficiency, and privacy of real training examples. Nevertheless, existing approaches for synthetic text generation are mostly heuristics and cannot generate human-readable text without compromising the privacy of real data, or provide performance guarantees for training Large Language Models (LLMs). In this work, we propose the first theoretically rigorous approach for generating synthetic human-readable text that provides convergence, performance, and privacy guarantees for fine-tuning LLMs on a target task. To do so, we le"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.17607","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/2502.17607/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:17:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"TwYvPt9L7ntBvcXaAWelf0LN4pcB4PL20/2lXNM2KBdW/DgKS4JmwWaV76F4T2MJkEzulIPn0W7vzWYJQBy+Ag==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T08:21:25.114724Z"},"content_sha256":"76d5a1806f2cc931e9d62d41dee05de6be75189124c176ba8b9c9a6aa50944b4","schema_version":"1.0","event_id":"sha256:76d5a1806f2cc931e9d62d41dee05de6be75189124c176ba8b9c9a6aa50944b4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TKXRKTDGZR47ZXPHD3W6LQS6BA/bundle.json","state_url":"https://pith.science/pith/TKXRKTDGZR47ZXPHD3W6LQS6BA/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TKXRKTDGZR47ZXPHD3W6LQS6BA/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-09T08:21:25Z","links":{"resolver":"https://pith.science/pith/TKXRKTDGZR47ZXPHD3W6LQS6BA","bundle":"https://pith.science/pith/TKXRKTDGZR47ZXPHD3W6LQS6BA/bundle.json","state":"https://pith.science/pith/TKXRKTDGZR47ZXPHD3W6LQS6BA/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TKXRKTDGZR47ZXPHD3W6LQS6BA/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:TKXRKTDGZR47ZXPHD3W6LQS6BA","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":"69f919e61d5758f0313b1569b779a2a294a215052094eda0262119001c4fb2c3","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T19:49:15Z","title_canon_sha256":"295bb1c7fd395e0560eea7246249456725b1fae5bb15b3bcf1a2b4ffc6c78151"},"schema_version":"1.0","source":{"id":"2502.17607","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.17607","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"arxiv_version","alias_value":"2502.17607v2","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.17607","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"pith_short_12","alias_value":"TKXRKTDGZR47","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"pith_short_16","alias_value":"TKXRKTDGZR47ZXPH","created_at":"2026-07-05T11:17:41Z"},{"alias_kind":"pith_short_8","alias_value":"TKXRKTDG","created_at":"2026-07-05T11:17:41Z"}],"graph_snapshots":[{"event_id":"sha256:76d5a1806f2cc931e9d62d41dee05de6be75189124c176ba8b9c9a6aa50944b4","target":"graph","created_at":"2026-07-05T11:17:41Z","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/2502.17607/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Synthetic data has the potential to improve the performance, training efficiency, and privacy of real training examples. Nevertheless, existing approaches for synthetic text generation are mostly heuristics and cannot generate human-readable text without compromising the privacy of real data, or provide performance guarantees for training Large Language Models (LLMs). In this work, we propose the first theoretically rigorous approach for generating synthetic human-readable text that provides convergence, performance, and privacy guarantees for fine-tuning LLMs on a target task. To do so, we le","authors_text":"Baharan Mirzasoleiman, Dang Nguyen, Meisam Razaviyayn, MohammadHossein Bateni, Vahab Mirrokni, Zeman Li","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T19:49:15Z","title":"Synthetic Text Generation for Training Large Language Models via Gradient Matching"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.17607","kind":"arxiv","version":2},"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":"sha256:261942f2c4e67dcbe5c03cb3933c2c204ce8173c7bdd4eedd2ab4164a9000655","target":"record","created_at":"2026-07-05T11:17:41Z","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":"69f919e61d5758f0313b1569b779a2a294a215052094eda0262119001c4fb2c3","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-24T19:49:15Z","title_canon_sha256":"295bb1c7fd395e0560eea7246249456725b1fae5bb15b3bcf1a2b4ffc6c78151"},"schema_version":"1.0","source":{"id":"2502.17607","kind":"arxiv","version":2}},"canonical_sha256":"9aaf154c66cc79fcdde71eede5c25e0816bc694990babb0b40ab124c16448834","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9aaf154c66cc79fcdde71eede5c25e0816bc694990babb0b40ab124c16448834","first_computed_at":"2026-07-05T11:17:41.927634Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:17:41.927634Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"VAypGMiD842rZ6NOULaZ/ZTxcqgrQD8SpqwNA7VRuU+0xKl9xOrJkpBOMNVY78yVQAUKVgjFfl2/+TfmKx2TDA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:17:41.928156Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.17607","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:261942f2c4e67dcbe5c03cb3933c2c204ce8173c7bdd4eedd2ab4164a9000655","sha256:76d5a1806f2cc931e9d62d41dee05de6be75189124c176ba8b9c9a6aa50944b4"],"state_sha256":"c798568f544a23a58e6df0048cf92dfbb7f06990542a942f1b64d1a93c43abb9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9FHARu6+ADzpJR+RxHKQ6uSkwGXa+KdrP8Zvnpo/Uzcat9J5fXZuOb+IGlH7DWM7Lritu4CVnZyl7JVuHWIkBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T08:21:25.120878Z","bundle_sha256":"503db8748feac64ffb98aeaf4b87a86a8ae0f23b948476b749058bcc76aba0a3"}}