{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2017:N672T5PKFSHVYHVOUHCFSVLRM3","short_pith_number":"pith:N672T5PK","canonical_record":{"source":{"id":"1705.07878","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-05-22T17:42:15Z","cross_cats_sorted":["cs.DC","cs.NE"],"title_canon_sha256":"0ca75976f56ec846cc8fc3f04070945df2f541780e91cb83abc34bf5bef52e43","abstract_canon_sha256":"ad15118648d3ba3858517b72b4309eb7c61661e66fd8f0b8ea21a9066123d7f7"},"schema_version":"1.0"},"canonical_sha256":"6fbfa9f5ea2c8f5c1eaea1c459557166fed2cc0694755fef4aaab9ba89c051a7","source":{"kind":"arxiv","id":"1705.07878","version":6},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1705.07878","created_at":"2026-05-18T00:27:04Z"},{"alias_kind":"arxiv_version","alias_value":"1705.07878v6","created_at":"2026-05-18T00:27:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1705.07878","created_at":"2026-05-18T00:27:04Z"},{"alias_kind":"pith_short_12","alias_value":"N672T5PKFSHV","created_at":"2026-05-18T12:31:31Z"},{"alias_kind":"pith_short_16","alias_value":"N672T5PKFSHVYHVO","created_at":"2026-05-18T12:31:31Z"},{"alias_kind":"pith_short_8","alias_value":"N672T5PK","created_at":"2026-05-18T12:31:31Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2017:N672T5PKFSHVYHVOUHCFSVLRM3","target":"record","payload":{"canonical_record":{"source":{"id":"1705.07878","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-05-22T17:42:15Z","cross_cats_sorted":["cs.DC","cs.NE"],"title_canon_sha256":"0ca75976f56ec846cc8fc3f04070945df2f541780e91cb83abc34bf5bef52e43","abstract_canon_sha256":"ad15118648d3ba3858517b72b4309eb7c61661e66fd8f0b8ea21a9066123d7f7"},"schema_version":"1.0"},"canonical_sha256":"6fbfa9f5ea2c8f5c1eaea1c459557166fed2cc0694755fef4aaab9ba89c051a7","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:27:04.651938Z","signature_b64":"kb/IXFN1hDUA/+m8kiy3Cw4mK5u/lDBkoDZj3jHNBRx24b1ciNR6w9xp/m1XiSwq6lxoSh/NIKM+ioAKU+40Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6fbfa9f5ea2c8f5c1eaea1c459557166fed2cc0694755fef4aaab9ba89c051a7","last_reissued_at":"2026-05-18T00:27:04.651423Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:27:04.651423Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1705.07878","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":"2026-05-18T00:27:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Qm7DKVUqAbv6iS1gwISRCCUmSc5dDknuGzXFc3FYJ2E935bNQML6qwVdRPEgl5Dc/SdzPDl94f8m1HLxvQpTDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T22:41:30.830997Z"},"content_sha256":"86e8ceb38ecf0b991de2eddb96e534e82cad1e9b0cdc4ae8f4c4a57eca253a2e","schema_version":"1.0","event_id":"sha256:86e8ceb38ecf0b991de2eddb96e534e82cad1e9b0cdc4ae8f4c4a57eca253a2e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2017:N672T5PKFSHVYHVOUHCFSVLRM3","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"TernGrad: Ternary Gradients to Reduce Communication in Distributed Deep Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.DC","cs.NE"],"primary_cat":"cs.LG","authors_text":"Chunpeng Wu, Cong Xu, Feng Yan, Hai Li, Wei Wen, Yandan Wang, Yiran Chen","submitted_at":"2017-05-22T17:42:15Z","abstract_excerpt":"High network communication cost for synchronizing gradients and parameters is the well-known bottleneck of distributed training. In this work, we propose TernGrad that uses ternary gradients to accelerate distributed deep learning in data parallelism. Our approach requires only three numerical levels {-1,0,1}, which can aggressively reduce the communication time. We mathematically prove the convergence of TernGrad under the assumption of a bound on gradients. Guided by the bound, we propose layer-wise ternarizing and gradient clipping to improve its convergence. Our experiments show that apply"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1705.07878","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":""},"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":"2026-05-18T00:27:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ftamNQ3LJ5UY1/+e5LBGSDMGKbANNaCH7NyMrq0t7BNOILAB3YVx31O49VVbW1tWJtTYsLA3+gY/hfCa+jBhDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T22:41:30.831714Z"},"content_sha256":"ea906e0d3d55b3c2b8301759ee4b919e655dd7f773491dfcb642f6318f522915","schema_version":"1.0","event_id":"sha256:ea906e0d3d55b3c2b8301759ee4b919e655dd7f773491dfcb642f6318f522915"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/N672T5PKFSHVYHVOUHCFSVLRM3/bundle.json","state_url":"https://pith.science/pith/N672T5PKFSHVYHVOUHCFSVLRM3/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/N672T5PKFSHVYHVOUHCFSVLRM3/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":"2026-06-07T22:41:30Z","links":{"resolver":"https://pith.science/pith/N672T5PKFSHVYHVOUHCFSVLRM3","bundle":"https://pith.science/pith/N672T5PKFSHVYHVOUHCFSVLRM3/bundle.json","state":"https://pith.science/pith/N672T5PKFSHVYHVOUHCFSVLRM3/state.json","well_known_bundle":"https://pith.science/.well-known/pith/N672T5PKFSHVYHVOUHCFSVLRM3/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2017:N672T5PKFSHVYHVOUHCFSVLRM3","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":"ad15118648d3ba3858517b72b4309eb7c61661e66fd8f0b8ea21a9066123d7f7","cross_cats_sorted":["cs.DC","cs.NE"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-05-22T17:42:15Z","title_canon_sha256":"0ca75976f56ec846cc8fc3f04070945df2f541780e91cb83abc34bf5bef52e43"},"schema_version":"1.0","source":{"id":"1705.07878","kind":"arxiv","version":6}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1705.07878","created_at":"2026-05-18T00:27:04Z"},{"alias_kind":"arxiv_version","alias_value":"1705.07878v6","created_at":"2026-05-18T00:27:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1705.07878","created_at":"2026-05-18T00:27:04Z"},{"alias_kind":"pith_short_12","alias_value":"N672T5PKFSHV","created_at":"2026-05-18T12:31:31Z"},{"alias_kind":"pith_short_16","alias_value":"N672T5PKFSHVYHVO","created_at":"2026-05-18T12:31:31Z"},{"alias_kind":"pith_short_8","alias_value":"N672T5PK","created_at":"2026-05-18T12:31:31Z"}],"graph_snapshots":[{"event_id":"sha256:ea906e0d3d55b3c2b8301759ee4b919e655dd7f773491dfcb642f6318f522915","target":"graph","created_at":"2026-05-18T00:27:04Z","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":"High network communication cost for synchronizing gradients and parameters is the well-known bottleneck of distributed training. In this work, we propose TernGrad that uses ternary gradients to accelerate distributed deep learning in data parallelism. Our approach requires only three numerical levels {-1,0,1}, which can aggressively reduce the communication time. We mathematically prove the convergence of TernGrad under the assumption of a bound on gradients. Guided by the bound, we propose layer-wise ternarizing and gradient clipping to improve its convergence. Our experiments show that apply","authors_text":"Chunpeng Wu, Cong Xu, Feng Yan, Hai Li, Wei Wen, Yandan Wang, Yiran Chen","cross_cats":["cs.DC","cs.NE"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-05-22T17:42:15Z","title":"TernGrad: Ternary Gradients to Reduce Communication in Distributed Deep Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1705.07878","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":"sha256:86e8ceb38ecf0b991de2eddb96e534e82cad1e9b0cdc4ae8f4c4a57eca253a2e","target":"record","created_at":"2026-05-18T00:27:04Z","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":"ad15118648d3ba3858517b72b4309eb7c61661e66fd8f0b8ea21a9066123d7f7","cross_cats_sorted":["cs.DC","cs.NE"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-05-22T17:42:15Z","title_canon_sha256":"0ca75976f56ec846cc8fc3f04070945df2f541780e91cb83abc34bf5bef52e43"},"schema_version":"1.0","source":{"id":"1705.07878","kind":"arxiv","version":6}},"canonical_sha256":"6fbfa9f5ea2c8f5c1eaea1c459557166fed2cc0694755fef4aaab9ba89c051a7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6fbfa9f5ea2c8f5c1eaea1c459557166fed2cc0694755fef4aaab9ba89c051a7","first_computed_at":"2026-05-18T00:27:04.651423Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T00:27:04.651423Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"kb/IXFN1hDUA/+m8kiy3Cw4mK5u/lDBkoDZj3jHNBRx24b1ciNR6w9xp/m1XiSwq6lxoSh/NIKM+ioAKU+40Cw==","signature_status":"signed_v1","signed_at":"2026-05-18T00:27:04.651938Z","signed_message":"canonical_sha256_bytes"},"source_id":"1705.07878","source_kind":"arxiv","source_version":6}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:86e8ceb38ecf0b991de2eddb96e534e82cad1e9b0cdc4ae8f4c4a57eca253a2e","sha256:ea906e0d3d55b3c2b8301759ee4b919e655dd7f773491dfcb642f6318f522915"],"state_sha256":"7a8b47acfc1ae5aa07bdc22f02487bbaa99d517320f6b8967388226fc06ff84a"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rCsoXMNnwt24yFf6zRV7ulM7CDomB1IndCfRjvuA3bR7UGllfOLXulQAs7EkFNEC7fQuL1sSpdRr62GXQ6RhCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-07T22:41:30.836026Z","bundle_sha256":"701e6d4dcf96d1cdea79c72482710aa6a4425d7a1e9cf94ecf83c6022b0f3f04"}}