{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2017:6KBFRPFV3DDRIGHRMIKZSSROED","short_pith_number":"pith:6KBFRPFV","canonical_record":{"source":{"id":"1710.06963","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-10-18T23:46:57Z","cross_cats_sorted":[],"title_canon_sha256":"80339ee51a64ef234c6665554bce3729842336f878d1833e6da762f5c4707c24","abstract_canon_sha256":"bbc393b3fae6e71179131380b7ce262db432f59239dc60f7422a95b636314b64"},"schema_version":"1.0"},"canonical_sha256":"f28258bcb5d8c71418f16215994a2e20e210bd5a554732260b8bb4d6dd73ddef","source":{"kind":"arxiv","id":"1710.06963","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1710.06963","created_at":"2026-05-18T00:22:39Z"},{"alias_kind":"arxiv_version","alias_value":"1710.06963v3","created_at":"2026-05-18T00:22:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1710.06963","created_at":"2026-05-18T00:22:39Z"},{"alias_kind":"pith_short_12","alias_value":"6KBFRPFV3DDR","created_at":"2026-05-18T12:31:03Z"},{"alias_kind":"pith_short_16","alias_value":"6KBFRPFV3DDRIGHR","created_at":"2026-05-18T12:31:03Z"},{"alias_kind":"pith_short_8","alias_value":"6KBFRPFV","created_at":"2026-05-18T12:31:03Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2017:6KBFRPFV3DDRIGHRMIKZSSROED","target":"record","payload":{"canonical_record":{"source":{"id":"1710.06963","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-10-18T23:46:57Z","cross_cats_sorted":[],"title_canon_sha256":"80339ee51a64ef234c6665554bce3729842336f878d1833e6da762f5c4707c24","abstract_canon_sha256":"bbc393b3fae6e71179131380b7ce262db432f59239dc60f7422a95b636314b64"},"schema_version":"1.0"},"canonical_sha256":"f28258bcb5d8c71418f16215994a2e20e210bd5a554732260b8bb4d6dd73ddef","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:22:39.379792Z","signature_b64":"peJoPyFXZpmPmjNKEGGuQW6icMtcXzjCUL75GW03FUHoBTwK6/gCSjcITAdnuugbOawasfvuEOP2BjHepz1GBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f28258bcb5d8c71418f16215994a2e20e210bd5a554732260b8bb4d6dd73ddef","last_reissued_at":"2026-05-18T00:22:39.379419Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:22:39.379419Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1710.06963","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T00:22:39Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Zqfgw1wblCLdTPKmjyjfISEVv2NXyaESmy5XiVd2tPRBeHAyNCwNfDEKz2ZqWnCSKG99/s1lZV1IajR6CUEzAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-03T03:34:11.517002Z"},"content_sha256":"33540184e3d539d1868598ceecbdf3a9e4c1df804f9b093752418087a199b555","schema_version":"1.0","event_id":"sha256:33540184e3d539d1868598ceecbdf3a9e4c1df804f9b093752418087a199b555"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2017:6KBFRPFV3DDRIGHRMIKZSSROED","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Learning Differentially Private Recurrent Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Daniel Ramage, H. Brendan McMahan, Kunal Talwar, Li Zhang","submitted_at":"2017-10-18T23:46:57Z","abstract_excerpt":"We demonstrate that it is possible to train large recurrent language models with user-level differential privacy guarantees with only a negligible cost in predictive accuracy. Our work builds on recent advances in the training of deep networks on user-partitioned data and privacy accounting for stochastic gradient descent. In particular, we add user-level privacy protection to the federated averaging algorithm, which makes \"large step\" updates from user-level data. Our work demonstrates that given a dataset with a sufficiently large number of users (a requirement easily met by even small inter"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1710.06963","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"2026-05-18T00:22:39Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vlHpKGQSzR52qCDGgq9HHir7NbGA4S89LawFs8S6PKWouWuJ5yR2YI+PLdVQULFNnHxNHI5TKBk3EvgCdOdwAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-03T03:34:11.517354Z"},"content_sha256":"e7d6179cf935d45cb6553103802d8b179d4ce0c72774befa36f7a3d68ce69f7b","schema_version":"1.0","event_id":"sha256:e7d6179cf935d45cb6553103802d8b179d4ce0c72774befa36f7a3d68ce69f7b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/6KBFRPFV3DDRIGHRMIKZSSROED/bundle.json","state_url":"https://pith.science/pith/6KBFRPFV3DDRIGHRMIKZSSROED/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/6KBFRPFV3DDRIGHRMIKZSSROED/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":"2026-06-03T03:34:11Z","links":{"resolver":"https://pith.science/pith/6KBFRPFV3DDRIGHRMIKZSSROED","bundle":"https://pith.science/pith/6KBFRPFV3DDRIGHRMIKZSSROED/bundle.json","state":"https://pith.science/pith/6KBFRPFV3DDRIGHRMIKZSSROED/state.json","well_known_bundle":"https://pith.science/.well-known/pith/6KBFRPFV3DDRIGHRMIKZSSROED/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2017:6KBFRPFV3DDRIGHRMIKZSSROED","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":"bbc393b3fae6e71179131380b7ce262db432f59239dc60f7422a95b636314b64","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-10-18T23:46:57Z","title_canon_sha256":"80339ee51a64ef234c6665554bce3729842336f878d1833e6da762f5c4707c24"},"schema_version":"1.0","source":{"id":"1710.06963","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1710.06963","created_at":"2026-05-18T00:22:39Z"},{"alias_kind":"arxiv_version","alias_value":"1710.06963v3","created_at":"2026-05-18T00:22:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1710.06963","created_at":"2026-05-18T00:22:39Z"},{"alias_kind":"pith_short_12","alias_value":"6KBFRPFV3DDR","created_at":"2026-05-18T12:31:03Z"},{"alias_kind":"pith_short_16","alias_value":"6KBFRPFV3DDRIGHR","created_at":"2026-05-18T12:31:03Z"},{"alias_kind":"pith_short_8","alias_value":"6KBFRPFV","created_at":"2026-05-18T12:31:03Z"}],"graph_snapshots":[{"event_id":"sha256:e7d6179cf935d45cb6553103802d8b179d4ce0c72774befa36f7a3d68ce69f7b","target":"graph","created_at":"2026-05-18T00:22:39Z","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"},"paper":{"abstract_excerpt":"We demonstrate that it is possible to train large recurrent language models with user-level differential privacy guarantees with only a negligible cost in predictive accuracy. Our work builds on recent advances in the training of deep networks on user-partitioned data and privacy accounting for stochastic gradient descent. In particular, we add user-level privacy protection to the federated averaging algorithm, which makes \"large step\" updates from user-level data. Our work demonstrates that given a dataset with a sufficiently large number of users (a requirement easily met by even small inter","authors_text":"Daniel Ramage, H. Brendan McMahan, Kunal Talwar, Li Zhang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-10-18T23:46:57Z","title":"Learning Differentially Private Recurrent Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1710.06963","kind":"arxiv","version":3},"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":"sha256:33540184e3d539d1868598ceecbdf3a9e4c1df804f9b093752418087a199b555","target":"record","created_at":"2026-05-18T00:22:39Z","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":"bbc393b3fae6e71179131380b7ce262db432f59239dc60f7422a95b636314b64","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-10-18T23:46:57Z","title_canon_sha256":"80339ee51a64ef234c6665554bce3729842336f878d1833e6da762f5c4707c24"},"schema_version":"1.0","source":{"id":"1710.06963","kind":"arxiv","version":3}},"canonical_sha256":"f28258bcb5d8c71418f16215994a2e20e210bd5a554732260b8bb4d6dd73ddef","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f28258bcb5d8c71418f16215994a2e20e210bd5a554732260b8bb4d6dd73ddef","first_computed_at":"2026-05-18T00:22:39.379419Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T00:22:39.379419Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"peJoPyFXZpmPmjNKEGGuQW6icMtcXzjCUL75GW03FUHoBTwK6/gCSjcITAdnuugbOawasfvuEOP2BjHepz1GBg==","signature_status":"signed_v1","signed_at":"2026-05-18T00:22:39.379792Z","signed_message":"canonical_sha256_bytes"},"source_id":"1710.06963","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:33540184e3d539d1868598ceecbdf3a9e4c1df804f9b093752418087a199b555","sha256:e7d6179cf935d45cb6553103802d8b179d4ce0c72774befa36f7a3d68ce69f7b"],"state_sha256":"fbb15b5124564f5aa8cb07abb0d51c9223582ebf294ad304bd19ce1236093172"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"580XtX9Ucs07f5vVhEWmTSwL441IkBYuJexW4xGOvyCoynLw5N4HG8r8rnQ8O7gSRlGGBIKHROa3BdyqthlHBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-03T03:34:11.519571Z","bundle_sha256":"f910b82898782216d372f170202893235b25d3f42274b5a105f2db2de770a9c2"}}