{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2018:6PIYLRGVFFU3AOQPEY7OQJNTWZ","short_pith_number":"pith:6PIYLRGV","canonical_record":{"source":{"id":"1808.08859","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2018-08-27T14:19:18Z","cross_cats_sorted":[],"title_canon_sha256":"37bad7b766ca3896cc551576a41175719b3dd1f26e4cefcfdce8f0b2216a019e","abstract_canon_sha256":"576452dc7f79a2b7be7ec2256a28a65a4cd2d3b6ef62d09facbccd72b53e9d7d"},"schema_version":"1.0"},"canonical_sha256":"f3d185c4d52969b03a0f263ee825b3b66735c7ce33f78113ebf2d2b0a2edd9ba","source":{"kind":"arxiv","id":"1808.08859","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1808.08859","created_at":"2026-05-18T00:05:45Z"},{"alias_kind":"arxiv_version","alias_value":"1808.08859v2","created_at":"2026-05-18T00:05:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1808.08859","created_at":"2026-05-18T00:05:45Z"},{"alias_kind":"pith_short_12","alias_value":"6PIYLRGVFFU3","created_at":"2026-05-18T12:32:11Z"},{"alias_kind":"pith_short_16","alias_value":"6PIYLRGVFFU3AOQP","created_at":"2026-05-18T12:32:11Z"},{"alias_kind":"pith_short_8","alias_value":"6PIYLRGV","created_at":"2026-05-18T12:32:11Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2018:6PIYLRGVFFU3AOQPEY7OQJNTWZ","target":"record","payload":{"canonical_record":{"source":{"id":"1808.08859","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2018-08-27T14:19:18Z","cross_cats_sorted":[],"title_canon_sha256":"37bad7b766ca3896cc551576a41175719b3dd1f26e4cefcfdce8f0b2216a019e","abstract_canon_sha256":"576452dc7f79a2b7be7ec2256a28a65a4cd2d3b6ef62d09facbccd72b53e9d7d"},"schema_version":"1.0"},"canonical_sha256":"f3d185c4d52969b03a0f263ee825b3b66735c7ce33f78113ebf2d2b0a2edd9ba","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:05:45.052141Z","signature_b64":"8ffScbfDHts5ZGT8N3U1u+6qBFxiPsts3gqGFezdqyD59owKZmYrVyUCsgJ7mJsVDhMq8EfpCo9KB4MmSKkkDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f3d185c4d52969b03a0f263ee825b3b66735c7ce33f78113ebf2d2b0a2edd9ba","last_reissued_at":"2026-05-18T00:05:45.051395Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:05:45.051395Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1808.08859","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":"2026-05-18T00:05:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Q+8Om616hEakAx9U8BPMGY16QUtQct5BT+82hGt3c/Txsd8MBGcYj9SIGlq5yD5+8GkpWjrfdP5dod1nfRaMCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T21:33:57.282011Z"},"content_sha256":"7a2b2b57e8c46ce33362c135ba5ddb906b9fcab34ee14880ae68e3ccfea6ff46","schema_version":"1.0","event_id":"sha256:7a2b2b57e8c46ce33362c135ba5ddb906b9fcab34ee14880ae68e3ccfea6ff46"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2018:6PIYLRGVFFU3AOQPEY7OQJNTWZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Accelerating Asynchronous Stochastic Gradient Descent for Neural Machine Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alham Fikri Aji, Kenneth Heafield, Marcin Junczys-Dowmunt, Nikolay Bogoychev","submitted_at":"2018-08-27T14:19:18Z","abstract_excerpt":"In order to extract the best possible performance from asynchronous stochastic gradient descent one must increase the mini-batch size and scale the learning rate accordingly. In order to achieve further speedup we introduce a technique that delays gradient updates effectively increasing the mini-batch size. Unfortunately with the increase of mini-batch size we worsen the stale gradient problem in asynchronous stochastic gradient descent (SGD) which makes the model convergence poor. We introduce local optimizers which mitigate the stale gradient problem and together with fine tuning our momentu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1808.08859","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":""},"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":"2026-05-18T00:05:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"CiiRumUZMuW/CEuDsd0NwYJR0tPNfz+F/vi68+8evZXTHsJLZbrT8XL/sz+X33FqL/rHcp4jBBIdiaFru+ZPCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T21:33:57.282650Z"},"content_sha256":"468c41b1643ffcc6337baed71f94ee59ab06c823f265f63cbe538bc3dfc7c1cc","schema_version":"1.0","event_id":"sha256:468c41b1643ffcc6337baed71f94ee59ab06c823f265f63cbe538bc3dfc7c1cc"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/6PIYLRGVFFU3AOQPEY7OQJNTWZ/bundle.json","state_url":"https://pith.science/pith/6PIYLRGVFFU3AOQPEY7OQJNTWZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/6PIYLRGVFFU3AOQPEY7OQJNTWZ/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":"2026-05-30T21:33:57Z","links":{"resolver":"https://pith.science/pith/6PIYLRGVFFU3AOQPEY7OQJNTWZ","bundle":"https://pith.science/pith/6PIYLRGVFFU3AOQPEY7OQJNTWZ/bundle.json","state":"https://pith.science/pith/6PIYLRGVFFU3AOQPEY7OQJNTWZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/6PIYLRGVFFU3AOQPEY7OQJNTWZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2018:6PIYLRGVFFU3AOQPEY7OQJNTWZ","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":"576452dc7f79a2b7be7ec2256a28a65a4cd2d3b6ef62d09facbccd72b53e9d7d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2018-08-27T14:19:18Z","title_canon_sha256":"37bad7b766ca3896cc551576a41175719b3dd1f26e4cefcfdce8f0b2216a019e"},"schema_version":"1.0","source":{"id":"1808.08859","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1808.08859","created_at":"2026-05-18T00:05:45Z"},{"alias_kind":"arxiv_version","alias_value":"1808.08859v2","created_at":"2026-05-18T00:05:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1808.08859","created_at":"2026-05-18T00:05:45Z"},{"alias_kind":"pith_short_12","alias_value":"6PIYLRGVFFU3","created_at":"2026-05-18T12:32:11Z"},{"alias_kind":"pith_short_16","alias_value":"6PIYLRGVFFU3AOQP","created_at":"2026-05-18T12:32:11Z"},{"alias_kind":"pith_short_8","alias_value":"6PIYLRGV","created_at":"2026-05-18T12:32:11Z"}],"graph_snapshots":[{"event_id":"sha256:468c41b1643ffcc6337baed71f94ee59ab06c823f265f63cbe538bc3dfc7c1cc","target":"graph","created_at":"2026-05-18T00:05:45Z","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":"In order to extract the best possible performance from asynchronous stochastic gradient descent one must increase the mini-batch size and scale the learning rate accordingly. In order to achieve further speedup we introduce a technique that delays gradient updates effectively increasing the mini-batch size. Unfortunately with the increase of mini-batch size we worsen the stale gradient problem in asynchronous stochastic gradient descent (SGD) which makes the model convergence poor. We introduce local optimizers which mitigate the stale gradient problem and together with fine tuning our momentu","authors_text":"Alham Fikri Aji, Kenneth Heafield, Marcin Junczys-Dowmunt, Nikolay Bogoychev","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2018-08-27T14:19:18Z","title":"Accelerating Asynchronous Stochastic Gradient Descent for Neural Machine Translation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1808.08859","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":"sha256:7a2b2b57e8c46ce33362c135ba5ddb906b9fcab34ee14880ae68e3ccfea6ff46","target":"record","created_at":"2026-05-18T00:05:45Z","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":"576452dc7f79a2b7be7ec2256a28a65a4cd2d3b6ef62d09facbccd72b53e9d7d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2018-08-27T14:19:18Z","title_canon_sha256":"37bad7b766ca3896cc551576a41175719b3dd1f26e4cefcfdce8f0b2216a019e"},"schema_version":"1.0","source":{"id":"1808.08859","kind":"arxiv","version":2}},"canonical_sha256":"f3d185c4d52969b03a0f263ee825b3b66735c7ce33f78113ebf2d2b0a2edd9ba","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f3d185c4d52969b03a0f263ee825b3b66735c7ce33f78113ebf2d2b0a2edd9ba","first_computed_at":"2026-05-18T00:05:45.051395Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T00:05:45.051395Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"8ffScbfDHts5ZGT8N3U1u+6qBFxiPsts3gqGFezdqyD59owKZmYrVyUCsgJ7mJsVDhMq8EfpCo9KB4MmSKkkDw==","signature_status":"signed_v1","signed_at":"2026-05-18T00:05:45.052141Z","signed_message":"canonical_sha256_bytes"},"source_id":"1808.08859","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7a2b2b57e8c46ce33362c135ba5ddb906b9fcab34ee14880ae68e3ccfea6ff46","sha256:468c41b1643ffcc6337baed71f94ee59ab06c823f265f63cbe538bc3dfc7c1cc"],"state_sha256":"70e72ea5e24fc2260fb70a11f0fba95dab778966a1099c29aa484ddd533e0fc0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"NKz3gMfer47qMo+8a/HF0rchnxPcQikhbnxqhlOx8VRPII03qrtoLf8Q7hlzTyIodwJerN9NL8ArNdv6SpwDBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-30T21:33:57.285805Z","bundle_sha256":"9ae96a1751c8f93fe83237ab3e5825603da700b564fc06fb89d13a4777226763"}}