{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2018:X6C7HE6GW3MZ7Z3NDLH3MRE6VI","short_pith_number":"pith:X6C7HE6G","canonical_record":{"source":{"id":"1808.07217","kind":"arxiv","version":6},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2018-08-22T04:50:55Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"9406e39837368c7f5ed49513798826363b768aeffd47ad72a4ab6af72d48d655","abstract_canon_sha256":"0129bd50d66cbe7b561b5269da5d44840bfc61ba1cee2cd8c586e8fd5d499f19"},"schema_version":"1.0"},"canonical_sha256":"bf85f393c6b6d99fe76d1acfb6449eaa312766b32db4a95759d6b8ee9c74d611","source":{"kind":"arxiv","id":"1808.07217","version":6},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1808.07217","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"arxiv_version","alias_value":"1808.07217v6","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1808.07217","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"pith_short_12","alias_value":"X6C7HE6GW3MZ","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"pith_short_16","alias_value":"X6C7HE6GW3MZ7Z3N","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"pith_short_8","alias_value":"X6C7HE6G","created_at":"2026-07-05T00:40:53Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2018:X6C7HE6GW3MZ7Z3NDLH3MRE6VI","target":"record","payload":{"canonical_record":{"source":{"id":"1808.07217","kind":"arxiv","version":6},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2018-08-22T04:50:55Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"9406e39837368c7f5ed49513798826363b768aeffd47ad72a4ab6af72d48d655","abstract_canon_sha256":"0129bd50d66cbe7b561b5269da5d44840bfc61ba1cee2cd8c586e8fd5d499f19"},"schema_version":"1.0"},"canonical_sha256":"bf85f393c6b6d99fe76d1acfb6449eaa312766b32db4a95759d6b8ee9c74d611","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:40:53.188114Z","signature_b64":"glEz0gsypUZFSgy8Qc5iftJel5y9vm10w7SupUVctGjnNQ55XsHRaa4mqxN+Tn69UgtwaXv7SzBAM8sMYEW4BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bf85f393c6b6d99fe76d1acfb6449eaa312766b32db4a95759d6b8ee9c74d611","last_reissued_at":"2026-07-05T00:40:53.187609Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:40:53.187609Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1808.07217","source_version":6,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T00:40:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vBoNKo8nLzRLdXTVKv1e77f7TJSRryBrh2mIPydkhcBB5PY2VU4gCshKVNY+QPMWP4/diDbXczkEG5QvJp9GCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T01:31:59.799896Z"},"content_sha256":"3b3ad391c2b0ac39d4d0449ea766f7f480514611c0f17ea04baec13f2381e299","schema_version":"1.0","event_id":"sha256:3b3ad391c2b0ac39d4d0449ea766f7f480514611c0f17ea04baec13f2381e299"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2018:X6C7HE6GW3MZ7Z3NDLH3MRE6VI","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Don't Use Large Mini-Batches, Use Local SGD","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Kumar Kshitij Patel, Martin Jaggi, Sebastian U. Stich, Tao Lin","submitted_at":"2018-08-22T04:50:55Z","abstract_excerpt":"Mini-batch stochastic gradient methods (SGD) are state of the art for distributed training of deep neural networks. Drastic increases in the mini-batch sizes have lead to key efficiency and scalability gains in recent years. However, progress faces a major roadblock, as models trained with large batches often do not generalize well, i.e. they do not show good accuracy on new data. As a remedy, we propose a \\emph{post-local} SGD and show that it significantly improves the generalization performance compared to large-batch training on standard benchmarks while enjoying the same efficiency (time-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1808.07217","kind":"arxiv","version":6},"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/1808.07217/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":"2026-07-05T00:40:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y9hHBLjvuZDJ75YoW8ZbV/q1jgAuWh48AOeIrdRrlx19cmicMEHkTgXbkbDxR6cOoBrpaJN/hhdTWkC+v4UgDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T01:31:59.800492Z"},"content_sha256":"c5ec9515dfab5c2b97f238646294a93993fd90530af62fcd1cb5c24178cce9ab","schema_version":"1.0","event_id":"sha256:c5ec9515dfab5c2b97f238646294a93993fd90530af62fcd1cb5c24178cce9ab"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/X6C7HE6GW3MZ7Z3NDLH3MRE6VI/bundle.json","state_url":"https://pith.science/pith/X6C7HE6GW3MZ7Z3NDLH3MRE6VI/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/X6C7HE6GW3MZ7Z3NDLH3MRE6VI/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":"2026-08-07T01:31:59Z","links":{"resolver":"https://pith.science/pith/X6C7HE6GW3MZ7Z3NDLH3MRE6VI","bundle":"https://pith.science/pith/X6C7HE6GW3MZ7Z3NDLH3MRE6VI/bundle.json","state":"https://pith.science/pith/X6C7HE6GW3MZ7Z3NDLH3MRE6VI/state.json","well_known_bundle":"https://pith.science/.well-known/pith/X6C7HE6GW3MZ7Z3NDLH3MRE6VI/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2018:X6C7HE6GW3MZ7Z3NDLH3MRE6VI","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":"0129bd50d66cbe7b561b5269da5d44840bfc61ba1cee2cd8c586e8fd5d499f19","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2018-08-22T04:50:55Z","title_canon_sha256":"9406e39837368c7f5ed49513798826363b768aeffd47ad72a4ab6af72d48d655"},"schema_version":"1.0","source":{"id":"1808.07217","kind":"arxiv","version":6}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1808.07217","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"arxiv_version","alias_value":"1808.07217v6","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1808.07217","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"pith_short_12","alias_value":"X6C7HE6GW3MZ","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"pith_short_16","alias_value":"X6C7HE6GW3MZ7Z3N","created_at":"2026-07-05T00:40:53Z"},{"alias_kind":"pith_short_8","alias_value":"X6C7HE6G","created_at":"2026-07-05T00:40:53Z"}],"graph_snapshots":[{"event_id":"sha256:c5ec9515dfab5c2b97f238646294a93993fd90530af62fcd1cb5c24178cce9ab","target":"graph","created_at":"2026-07-05T00:40:53Z","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/1808.07217/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Mini-batch stochastic gradient methods (SGD) are state of the art for distributed training of deep neural networks. Drastic increases in the mini-batch sizes have lead to key efficiency and scalability gains in recent years. However, progress faces a major roadblock, as models trained with large batches often do not generalize well, i.e. they do not show good accuracy on new data. As a remedy, we propose a \\emph{post-local} SGD and show that it significantly improves the generalization performance compared to large-batch training on standard benchmarks while enjoying the same efficiency (time-","authors_text":"Kumar Kshitij Patel, Martin Jaggi, Sebastian U. Stich, Tao Lin","cross_cats":["stat.ML"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2018-08-22T04:50:55Z","title":"Don't Use Large Mini-Batches, Use Local SGD"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1808.07217","kind":"arxiv","version":6},"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":"sha256:3b3ad391c2b0ac39d4d0449ea766f7f480514611c0f17ea04baec13f2381e299","target":"record","created_at":"2026-07-05T00:40:53Z","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":"0129bd50d66cbe7b561b5269da5d44840bfc61ba1cee2cd8c586e8fd5d499f19","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2018-08-22T04:50:55Z","title_canon_sha256":"9406e39837368c7f5ed49513798826363b768aeffd47ad72a4ab6af72d48d655"},"schema_version":"1.0","source":{"id":"1808.07217","kind":"arxiv","version":6}},"canonical_sha256":"bf85f393c6b6d99fe76d1acfb6449eaa312766b32db4a95759d6b8ee9c74d611","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bf85f393c6b6d99fe76d1acfb6449eaa312766b32db4a95759d6b8ee9c74d611","first_computed_at":"2026-07-05T00:40:53.187609Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T00:40:53.187609Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"glEz0gsypUZFSgy8Qc5iftJel5y9vm10w7SupUVctGjnNQ55XsHRaa4mqxN+Tn69UgtwaXv7SzBAM8sMYEW4BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T00:40:53.188114Z","signed_message":"canonical_sha256_bytes"},"source_id":"1808.07217","source_kind":"arxiv","source_version":6}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3b3ad391c2b0ac39d4d0449ea766f7f480514611c0f17ea04baec13f2381e299","sha256:c5ec9515dfab5c2b97f238646294a93993fd90530af62fcd1cb5c24178cce9ab"],"state_sha256":"ea86badabff79c183fd5b942c770c34ecf5f7dfb0dc06e42f92c499830b1a453"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"dgL6BDB6knaRYtcC8XcLM9GLDLZL0XP1jYAGvCi4tYDjwqPmqWzkWQZOAv+tiRahxDI4a3/1hegn8asgN4ePCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T01:31:59.806476Z","bundle_sha256":"abc845a472d63e9535b595e4aefce5f34998cace0bedb2d75afb82dc90ff0af1"}}