{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2017:4W5T6IJYEXSSVTEEQU2MXK4DO4","short_pith_number":"pith:4W5T6IJY","schema_version":"1.0","canonical_sha256":"e5bb3f213825e52acc848534cbab83773a13596580c7c09eb4de64fd1bd0919a","source":{"kind":"arxiv","id":"1712.02679","version":1},"attestation_state":"computed","paper":{"title":"AdaComp : Adaptive Residual Gradient Compression for Data-Parallel Distributed Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Ankur Agrawal, Chia-Yu Chen, Daniel Brand, Jungwook Choi, Kailash Gopalakrishnan, Wei Zhang","submitted_at":"2017-12-07T15:44:19Z","abstract_excerpt":"Highly distributed training of Deep Neural Networks (DNNs) on future compute platforms (offering 100 of TeraOps/s of computational capacity) is expected to be severely communication constrained. To overcome this limitation, new gradient compression techniques are needed that are computationally friendly, applicable to a wide variety of layers seen in Deep Neural Networks and adaptable to variations in network architectures as well as their hyper-parameters. In this paper we introduce a novel technique - the Adaptive Residual Gradient Compression (AdaComp) scheme. AdaComp is based on localized "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"1712.02679","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-12-07T15:44:19Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"fa61259fb836363f43d33b895eb19c0f1e3f9e980786e38f025350093b0f99c3","abstract_canon_sha256":"11fa79846666f3ba2fb4b70cbfd1fa3c4c7ba1d97f19866127949b4187905ccd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:28:33.207539Z","signature_b64":"eUqh/6R+6tlw7LYNuDtlwZ1BK1+mK9KaGfWsGnatBZWSkhfDv75RBhRrkCdgkEXUu6MMsv4KZKce44xsEWldBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e5bb3f213825e52acc848534cbab83773a13596580c7c09eb4de64fd1bd0919a","last_reissued_at":"2026-05-18T00:28:33.206878Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:28:33.206878Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AdaComp : Adaptive Residual Gradient Compression for Data-Parallel Distributed Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Ankur Agrawal, Chia-Yu Chen, Daniel Brand, Jungwook Choi, Kailash Gopalakrishnan, Wei Zhang","submitted_at":"2017-12-07T15:44:19Z","abstract_excerpt":"Highly distributed training of Deep Neural Networks (DNNs) on future compute platforms (offering 100 of TeraOps/s of computational capacity) is expected to be severely communication constrained. To overcome this limitation, new gradient compression techniques are needed that are computationally friendly, applicable to a wide variety of layers seen in Deep Neural Networks and adaptable to variations in network architectures as well as their hyper-parameters. In this paper we introduce a novel technique - the Adaptive Residual Gradient Compression (AdaComp) scheme. AdaComp is based on localized "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1712.02679","kind":"arxiv","version":1},"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"},"aliases":[{"alias_kind":"arxiv","alias_value":"1712.02679","created_at":"2026-05-18T00:28:33.206983+00:00"},{"alias_kind":"arxiv_version","alias_value":"1712.02679v1","created_at":"2026-05-18T00:28:33.206983+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1712.02679","created_at":"2026-05-18T00:28:33.206983+00:00"},{"alias_kind":"pith_short_12","alias_value":"4W5T6IJYEXSS","created_at":"2026-05-18T12:31:00.734936+00:00"},{"alias_kind":"pith_short_16","alias_value":"4W5T6IJYEXSSVTEE","created_at":"2026-05-18T12:31:00.734936+00:00"},{"alias_kind":"pith_short_8","alias_value":"4W5T6IJY","created_at":"2026-05-18T12:31:00.734936+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4","json":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4.json","graph_json":"https://pith.science/api/pith-number/4W5T6IJYEXSSVTEEQU2MXK4DO4/graph.json","events_json":"https://pith.science/api/pith-number/4W5T6IJYEXSSVTEEQU2MXK4DO4/events.json","paper":"https://pith.science/paper/4W5T6IJY"},"agent_actions":{"view_html":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4","download_json":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4.json","view_paper":"https://pith.science/paper/4W5T6IJY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1712.02679&json=true","fetch_graph":"https://pith.science/api/pith-number/4W5T6IJYEXSSVTEEQU2MXK4DO4/graph.json","fetch_events":"https://pith.science/api/pith-number/4W5T6IJYEXSSVTEEQU2MXK4DO4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4/action/storage_attestation","attest_author":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4/action/author_attestation","sign_citation":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4/action/citation_signature","submit_replication":"https://pith.science/pith/4W5T6IJYEXSSVTEEQU2MXK4DO4/action/replication_record"}},"created_at":"2026-05-18T00:28:33.206983+00:00","updated_at":"2026-05-18T00:28:33.206983+00:00"}