{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:5QE7C4K3OR7CCQ3OR5HGQ7IKK6","short_pith_number":"pith:5QE7C4K3","canonical_record":{"source":{"id":"2306.00169","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-31T20:28:13Z","cross_cats_sorted":[],"title_canon_sha256":"0b7c66cbfaf9b629fbbbeb2406582bfe40e5b988f0ae8b010b279e3af84af731","abstract_canon_sha256":"10f3cf972d293af898560308d254236dcaaca9c4879e235977df898c8f945455"},"schema_version":"1.0"},"canonical_sha256":"ec09f1715b747e21436e8f4e687d0a57a3dc2720aa5763d597fc9d2156759f93","source":{"kind":"arxiv","id":"2306.00169","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2306.00169","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"arxiv_version","alias_value":"2306.00169v2","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.00169","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"pith_short_12","alias_value":"5QE7C4K3OR7C","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"pith_short_16","alias_value":"5QE7C4K3OR7CCQ3O","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"pith_short_8","alias_value":"5QE7C4K3","created_at":"2026-07-05T07:06:28Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:5QE7C4K3OR7CCQ3OR5HGQ7IKK6","target":"record","payload":{"canonical_record":{"source":{"id":"2306.00169","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-31T20:28:13Z","cross_cats_sorted":[],"title_canon_sha256":"0b7c66cbfaf9b629fbbbeb2406582bfe40e5b988f0ae8b010b279e3af84af731","abstract_canon_sha256":"10f3cf972d293af898560308d254236dcaaca9c4879e235977df898c8f945455"},"schema_version":"1.0"},"canonical_sha256":"ec09f1715b747e21436e8f4e687d0a57a3dc2720aa5763d597fc9d2156759f93","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:06:28.936292Z","signature_b64":"p+I30ckTEO52c+QX+VipPBUg/uO9lAj8hUM7tNeGqShtlH5LuvFpxxV98wwnFRlGyYR9PU2qURCoyqI6z7ABCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ec09f1715b747e21436e8f4e687d0a57a3dc2720aa5763d597fc9d2156759f93","last_reissued_at":"2026-07-05T07:06:28.935864Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:06:28.935864Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2306.00169","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":"2026-07-05T07:06:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4kA6SIOnBggvbg4hUZ/VcSGTH3O0NDLsF2De+rH0uSNwYVvwe7Bsd7/nAHbYB6dc/2sOg8DEW1VG/CS5BXbkAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-20T08:46:39.200205Z"},"content_sha256":"8792a00c99016766448dc198d71eab59590bec8cab44157d898623d8b7649b4f","schema_version":"1.0","event_id":"sha256:8792a00c99016766448dc198d71eab59590bec8cab44157d898623d8b7649b4f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:5QE7C4K3OR7CCQ3OR5HGQ7IKK6","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Inconsistency, Instability, and Generalization Gap of Deep Neural Network Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Rie Johnson, Tong Zhang","submitted_at":"2023-05-31T20:28:13Z","abstract_excerpt":"As deep neural networks are highly expressive, it is important to find solutions with small generalization gap (the difference between the performance on the training data and unseen data). Focusing on the stochastic nature of training, we first present a theoretical analysis in which the bound of generalization gap depends on what we call inconsistency and instability of model outputs, which can be estimated on unlabeled data. Our empirical study based on this analysis shows that instability and inconsistency are strongly predictive of generalization gap in various settings. In particular, ou"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.00169","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/2306.00169/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":"2026-07-05T07:06:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Hdk6bhA3S14AlWJKXEyBVXzuaEnI3Ux8DxBHvDvbOz51rhCPaIGq9+ek83/Jhzyx14B2wY49Fnl/GPMB8sbYDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-20T08:46:39.200877Z"},"content_sha256":"695d32696e438b063ed128d517beb40c05a57519534e443cbfb47962d000866a","schema_version":"1.0","event_id":"sha256:695d32696e438b063ed128d517beb40c05a57519534e443cbfb47962d000866a"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/5QE7C4K3OR7CCQ3OR5HGQ7IKK6/bundle.json","state_url":"https://pith.science/pith/5QE7C4K3OR7CCQ3OR5HGQ7IKK6/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/5QE7C4K3OR7CCQ3OR5HGQ7IKK6/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":"2026-08-20T08:46:39Z","links":{"resolver":"https://pith.science/pith/5QE7C4K3OR7CCQ3OR5HGQ7IKK6","bundle":"https://pith.science/pith/5QE7C4K3OR7CCQ3OR5HGQ7IKK6/bundle.json","state":"https://pith.science/pith/5QE7C4K3OR7CCQ3OR5HGQ7IKK6/state.json","well_known_bundle":"https://pith.science/.well-known/pith/5QE7C4K3OR7CCQ3OR5HGQ7IKK6/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:5QE7C4K3OR7CCQ3OR5HGQ7IKK6","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":"10f3cf972d293af898560308d254236dcaaca9c4879e235977df898c8f945455","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-31T20:28:13Z","title_canon_sha256":"0b7c66cbfaf9b629fbbbeb2406582bfe40e5b988f0ae8b010b279e3af84af731"},"schema_version":"1.0","source":{"id":"2306.00169","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2306.00169","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"arxiv_version","alias_value":"2306.00169v2","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.00169","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"pith_short_12","alias_value":"5QE7C4K3OR7C","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"pith_short_16","alias_value":"5QE7C4K3OR7CCQ3O","created_at":"2026-07-05T07:06:28Z"},{"alias_kind":"pith_short_8","alias_value":"5QE7C4K3","created_at":"2026-07-05T07:06:28Z"}],"graph_snapshots":[{"event_id":"sha256:695d32696e438b063ed128d517beb40c05a57519534e443cbfb47962d000866a","target":"graph","created_at":"2026-07-05T07:06:28Z","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/2306.00169/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"As deep neural networks are highly expressive, it is important to find solutions with small generalization gap (the difference between the performance on the training data and unseen data). Focusing on the stochastic nature of training, we first present a theoretical analysis in which the bound of generalization gap depends on what we call inconsistency and instability of model outputs, which can be estimated on unlabeled data. Our empirical study based on this analysis shows that instability and inconsistency are strongly predictive of generalization gap in various settings. In particular, ou","authors_text":"Rie Johnson, Tong Zhang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-31T20:28:13Z","title":"Inconsistency, Instability, and Generalization Gap of Deep Neural Network Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.00169","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":"sha256:8792a00c99016766448dc198d71eab59590bec8cab44157d898623d8b7649b4f","target":"record","created_at":"2026-07-05T07:06:28Z","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":"10f3cf972d293af898560308d254236dcaaca9c4879e235977df898c8f945455","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-31T20:28:13Z","title_canon_sha256":"0b7c66cbfaf9b629fbbbeb2406582bfe40e5b988f0ae8b010b279e3af84af731"},"schema_version":"1.0","source":{"id":"2306.00169","kind":"arxiv","version":2}},"canonical_sha256":"ec09f1715b747e21436e8f4e687d0a57a3dc2720aa5763d597fc9d2156759f93","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ec09f1715b747e21436e8f4e687d0a57a3dc2720aa5763d597fc9d2156759f93","first_computed_at":"2026-07-05T07:06:28.935864Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:06:28.935864Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"p+I30ckTEO52c+QX+VipPBUg/uO9lAj8hUM7tNeGqShtlH5LuvFpxxV98wwnFRlGyYR9PU2qURCoyqI6z7ABCQ==","signature_status":"signed_v1","signed_at":"2026-07-05T07:06:28.936292Z","signed_message":"canonical_sha256_bytes"},"source_id":"2306.00169","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:8792a00c99016766448dc198d71eab59590bec8cab44157d898623d8b7649b4f","sha256:695d32696e438b063ed128d517beb40c05a57519534e443cbfb47962d000866a"],"state_sha256":"4eaf99b9f121d72b9d0c5897621bb03f6c36205bdf9ea8bccd9d86bb3f756ea2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xNN4dRQofic10xkKOVFSOSThJoM7PTM6wu8FLSTuHTy6m9gDfkp4nQMrMOyH9G6S4QRx7kl87usnUDu6qT33CQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-20T08:46:39.206359Z","bundle_sha256":"cc2c668ff62b9886bd15cc79188c1815a9a0922f4d3a541823ffc96ebf6f3000"}}