{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:WXNI57C2SJ24QK62DB7DTUREA3","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":"c94a0894340901feab0d2f9f97790630d767c95063f83d03d8884cacbb3e75b8","cross_cats_sorted":["cs.DC"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-11-07T00:18:43Z","title_canon_sha256":"d0923d88f5513d1faa816d48831900ab94519a7541823483ad2d50c2c989a5f6"},"schema_version":"1.0","source":{"id":"2011.03641","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2011.03641","created_at":"2026-07-05T02:23:17Z"},{"alias_kind":"arxiv_version","alias_value":"2011.03641v3","created_at":"2026-07-05T02:23:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2011.03641","created_at":"2026-07-05T02:23:17Z"},{"alias_kind":"pith_short_12","alias_value":"WXNI57C2SJ24","created_at":"2026-07-05T02:23:17Z"},{"alias_kind":"pith_short_16","alias_value":"WXNI57C2SJ24QK62","created_at":"2026-07-05T02:23:17Z"},{"alias_kind":"pith_short_8","alias_value":"WXNI57C2","created_at":"2026-07-05T02:23:17Z"}],"graph_snapshots":[{"event_id":"sha256:d9f9e54249fb10e747ea64ca0c97cf7493e8d51e8605700d875e14d946a805b7","target":"graph","created_at":"2026-07-05T02:23:17Z","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/2011.03641/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent results in language understanding using neural networks have required training hardware of unprecedentedscale, with thousands of chips cooperating on a single training run. This paper presents techniques to scaleML models on the Google TPU Multipod, a mesh with 4096 TPU-v3 chips. We discuss model parallelism toovercome scaling limitations from the fixed batch size in data parallelism, communication/collective optimizations,distributed evaluation of training metrics, and host input processing scaling optimizations. These techniques aredemonstrated in both the TensorFlow and JAX programmi","authors_text":"Andy Swing, Anselm Levskaya, Blake Hechtman, Cliff Young, Dehao Chen, HyoukJoong Lee, James Bradbury, Mehmet Deveci, Naveen Kumar, Pankaj Kanwar, Sameer Kumar, Shibo Wang, Skye Wanderman-Milne, Steve Lacy, Tao Wang, Tayo Oguntebi, Yazhou Zu, Yuanzhong Xu, Yu Emma Wang","cross_cats":["cs.DC"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-11-07T00:18:43Z","title":"Exploring the limits of Concurrency in ML Training on Google TPUs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2011.03641","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":"sha256:53247cbd541c428f11856593e03172ff2ca6c296b9d5eca6357c0cf99c64cf93","target":"record","created_at":"2026-07-05T02:23:17Z","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":"c94a0894340901feab0d2f9f97790630d767c95063f83d03d8884cacbb3e75b8","cross_cats_sorted":["cs.DC"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-11-07T00:18:43Z","title_canon_sha256":"d0923d88f5513d1faa816d48831900ab94519a7541823483ad2d50c2c989a5f6"},"schema_version":"1.0","source":{"id":"2011.03641","kind":"arxiv","version":3}},"canonical_sha256":"b5da8efc5a9275c82bda187e39d22406cb4d4359bb8101b2b8200a6c524a5b91","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b5da8efc5a9275c82bda187e39d22406cb4d4359bb8101b2b8200a6c524a5b91","first_computed_at":"2026-07-05T02:23:17.585731Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T02:23:17.585731Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"WH/X9OGySgSl/rdymZSl6o/2yvUtLCwAy+puLg0CA0Unj6dfWcEt/CnOdrWT2D7Ju83v4gFKjylKBIiWnlj8Bw==","signature_status":"signed_v1","signed_at":"2026-07-05T02:23:17.586151Z","signed_message":"canonical_sha256_bytes"},"source_id":"2011.03641","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:53247cbd541c428f11856593e03172ff2ca6c296b9d5eca6357c0cf99c64cf93","sha256:d9f9e54249fb10e747ea64ca0c97cf7493e8d51e8605700d875e14d946a805b7"],"state_sha256":"80a949784e176f12b8f5d3014c3664ce2e363415b35a8037d8d6393ae5dd48db"}