{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:MICA4MANQOEZPM6W5NP6W7I6HM","short_pith_number":"pith:MICA4MAN","canonical_record":{"source":{"id":"2201.00971","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-01-04T04:23:38Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"c8cbdadad4aa3a8843ada03f66ccb32df0f0e054cf32bfd68b1542e5bb0608d1","abstract_canon_sha256":"7e11437a8fe9951e5791b37de39da23221c31ecd5bc7ff77647fcad84d2b950c"},"schema_version":"1.0"},"canonical_sha256":"62040e300d838997b3d6eb5feb7d1e3b2e82624b727a2ac1ac5be0b651d3c8f0","source":{"kind":"arxiv","id":"2201.00971","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2201.00971","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"arxiv_version","alias_value":"2201.00971v1","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2201.00971","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"pith_short_12","alias_value":"MICA4MANQOEZ","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"pith_short_16","alias_value":"MICA4MANQOEZPM6W","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"pith_short_8","alias_value":"MICA4MAN","created_at":"2026-07-05T03:45:14Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:MICA4MANQOEZPM6W5NP6W7I6HM","target":"record","payload":{"canonical_record":{"source":{"id":"2201.00971","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-01-04T04:23:38Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"c8cbdadad4aa3a8843ada03f66ccb32df0f0e054cf32bfd68b1542e5bb0608d1","abstract_canon_sha256":"7e11437a8fe9951e5791b37de39da23221c31ecd5bc7ff77647fcad84d2b950c"},"schema_version":"1.0"},"canonical_sha256":"62040e300d838997b3d6eb5feb7d1e3b2e82624b727a2ac1ac5be0b651d3c8f0","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:45:14.794139Z","signature_b64":"bS6+N3Jw47DekBVQ8ydPC3PgvLeObCiFAsvXButyhnEJNXKtb6STnJ6mnTPLASNIXp3ZMhoWCqMYe38yxWnpCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"62040e300d838997b3d6eb5feb7d1e3b2e82624b727a2ac1ac5be0b651d3c8f0","last_reissued_at":"2026-07-05T03:45:14.793605Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:45:14.793605Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2201.00971","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T03:45:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"IjbhnN1F3V6ItRvSJj7rHDrk+GLEkRS1P1jheCsPTxB/jARY2b3xb9FO51F4uMT6hm4HOH7aOks1z4LwotwWDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T14:34:36.059610Z"},"content_sha256":"15e953984424034c20a142f412024b121ef5e14075c583a662d94c45d22dc0c3","schema_version":"1.0","event_id":"sha256:15e953984424034c20a142f412024b121ef5e14075c583a662d94c45d22dc0c3"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:MICA4MANQOEZPM6W5NP6W7I6HM","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Submix: Practical Private Prediction for Large-Scale Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Antonio Ginart, Chuan Guo, James Zou, Laurens van der Maaten","submitted_at":"2022-01-04T04:23:38Z","abstract_excerpt":"Recent data-extraction attacks have exposed that language models can memorize some training samples verbatim. This is a vulnerability that can compromise the privacy of the model's training data. In this work, we introduce SubMix: a practical protocol for private next-token prediction designed to prevent privacy violations by language models that were fine-tuned on a private corpus after pre-training on a public corpus. We show that SubMix limits the leakage of information that is unique to any individual user in the private corpus via a relaxation of group differentially private prediction. I"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2201.00971","kind":"arxiv","version":1},"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/2201.00971/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":"2026-07-05T03:45:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HFPoJgX2Z7gfCG1SMosSYGtITrf9of37onezB46rOUswZHNRI+4Qlgoppx6+BNqWYPmkhkXJOHVnzNaU+ZorBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T14:34:36.060141Z"},"content_sha256":"f7a2162e8f9543a884fec66065eb1fe8776a932c077073dc11ca661e5cd31ee2","schema_version":"1.0","event_id":"sha256:f7a2162e8f9543a884fec66065eb1fe8776a932c077073dc11ca661e5cd31ee2"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/MICA4MANQOEZPM6W5NP6W7I6HM/bundle.json","state_url":"https://pith.science/pith/MICA4MANQOEZPM6W5NP6W7I6HM/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/MICA4MANQOEZPM6W5NP6W7I6HM/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":"2026-08-10T14:34:36Z","links":{"resolver":"https://pith.science/pith/MICA4MANQOEZPM6W5NP6W7I6HM","bundle":"https://pith.science/pith/MICA4MANQOEZPM6W5NP6W7I6HM/bundle.json","state":"https://pith.science/pith/MICA4MANQOEZPM6W5NP6W7I6HM/state.json","well_known_bundle":"https://pith.science/.well-known/pith/MICA4MANQOEZPM6W5NP6W7I6HM/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:MICA4MANQOEZPM6W5NP6W7I6HM","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":"7e11437a8fe9951e5791b37de39da23221c31ecd5bc7ff77647fcad84d2b950c","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-01-04T04:23:38Z","title_canon_sha256":"c8cbdadad4aa3a8843ada03f66ccb32df0f0e054cf32bfd68b1542e5bb0608d1"},"schema_version":"1.0","source":{"id":"2201.00971","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2201.00971","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"arxiv_version","alias_value":"2201.00971v1","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2201.00971","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"pith_short_12","alias_value":"MICA4MANQOEZ","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"pith_short_16","alias_value":"MICA4MANQOEZPM6W","created_at":"2026-07-05T03:45:14Z"},{"alias_kind":"pith_short_8","alias_value":"MICA4MAN","created_at":"2026-07-05T03:45:14Z"}],"graph_snapshots":[{"event_id":"sha256:f7a2162e8f9543a884fec66065eb1fe8776a932c077073dc11ca661e5cd31ee2","target":"graph","created_at":"2026-07-05T03:45:14Z","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/2201.00971/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent data-extraction attacks have exposed that language models can memorize some training samples verbatim. This is a vulnerability that can compromise the privacy of the model's training data. In this work, we introduce SubMix: a practical protocol for private next-token prediction designed to prevent privacy violations by language models that were fine-tuned on a private corpus after pre-training on a public corpus. We show that SubMix limits the leakage of information that is unique to any individual user in the private corpus via a relaxation of group differentially private prediction. I","authors_text":"Antonio Ginart, Chuan Guo, James Zou, Laurens van der Maaten","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-01-04T04:23:38Z","title":"Submix: Practical Private Prediction for Large-Scale Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2201.00971","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":"sha256:15e953984424034c20a142f412024b121ef5e14075c583a662d94c45d22dc0c3","target":"record","created_at":"2026-07-05T03:45:14Z","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":"7e11437a8fe9951e5791b37de39da23221c31ecd5bc7ff77647fcad84d2b950c","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-01-04T04:23:38Z","title_canon_sha256":"c8cbdadad4aa3a8843ada03f66ccb32df0f0e054cf32bfd68b1542e5bb0608d1"},"schema_version":"1.0","source":{"id":"2201.00971","kind":"arxiv","version":1}},"canonical_sha256":"62040e300d838997b3d6eb5feb7d1e3b2e82624b727a2ac1ac5be0b651d3c8f0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"62040e300d838997b3d6eb5feb7d1e3b2e82624b727a2ac1ac5be0b651d3c8f0","first_computed_at":"2026-07-05T03:45:14.793605Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T03:45:14.793605Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"bS6+N3Jw47DekBVQ8ydPC3PgvLeObCiFAsvXButyhnEJNXKtb6STnJ6mnTPLASNIXp3ZMhoWCqMYe38yxWnpCA==","signature_status":"signed_v1","signed_at":"2026-07-05T03:45:14.794139Z","signed_message":"canonical_sha256_bytes"},"source_id":"2201.00971","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:15e953984424034c20a142f412024b121ef5e14075c583a662d94c45d22dc0c3","sha256:f7a2162e8f9543a884fec66065eb1fe8776a932c077073dc11ca661e5cd31ee2"],"state_sha256":"c6ff29a85316a9008430d9fa255b231cbff9e832f5f337c25ee82fb8c7d96e4d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lL6Rnqhjz7ars+tqe5pNnjScGHqXniwf/umhQrvHAITvWuvqodpkzO4OjleFWzpLkyKXC0C8K8p0MFLUt7RCDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-10T14:34:36.070427Z","bundle_sha256":"639e7c4a30ccd79a2bc659d53de9afb677507c394afa9f1d698915e0e899ac97"}}