{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:SZ32ZXJZ5WEO3OCJNKVKHZAH4A","short_pith_number":"pith:SZ32ZXJZ","schema_version":"1.0","canonical_sha256":"9677acdd39ed88edb8496aaaa3e407e02c0cec12b8acd730cb7f7aad24e99de8","source":{"kind":"arxiv","id":"1804.03235","version":2},"attestation_state":"computed","paper":{"title":"Large scale distributed neural network training through online distillation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alexandre Passos, Gabriel Pereyra, Geoffrey E. Hinton, George E. Dahl, Robert Ormandi, Rohan Anil","submitted_at":"2018-04-09T20:56:03Z","abstract_excerpt":"Techniques such as ensembling and distillation promise model quality improvements when paired with almost any base model. However, due to increased test-time cost (for ensembles) and increased complexity of the training pipeline (for distillation), these techniques are challenging to use in industrial settings. In this paper we explore a variant of distillation which is relatively straightforward to use as it does not require a complicated multi-stage setup or many new hyperparameters. Our first claim is that online distillation enables us to use extra parallelism to fit very large datasets ab"},"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":"1804.03235","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2018-04-09T20:56:03Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"7958c90213517a3bda6f77fc4cdb280b493287a9d81d4cccea2d80835da4036f","abstract_canon_sha256":"097e11ec9ab7130cd98086be71d82fc13f286191a073c906253784ec0fb59855"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:28:39.925342Z","signature_b64":"vbM2NIHzcOJWUqtFOEFXrdXmXZI1hnPne1wCTvDbI88dGubF3CUE7r4hwwEem+Utf1AkhxDvs5QvPIkHZfBpBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9677acdd39ed88edb8496aaaa3e407e02c0cec12b8acd730cb7f7aad24e99de8","last_reissued_at":"2026-07-05T01:28:39.924832Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:28:39.924832Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large scale distributed neural network training through online distillation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alexandre Passos, Gabriel Pereyra, Geoffrey E. Hinton, George E. Dahl, Robert Ormandi, Rohan Anil","submitted_at":"2018-04-09T20:56:03Z","abstract_excerpt":"Techniques such as ensembling and distillation promise model quality improvements when paired with almost any base model. However, due to increased test-time cost (for ensembles) and increased complexity of the training pipeline (for distillation), these techniques are challenging to use in industrial settings. In this paper we explore a variant of distillation which is relatively straightforward to use as it does not require a complicated multi-stage setup or many new hyperparameters. Our first claim is that online distillation enables us to use extra parallelism to fit very large datasets ab"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1804.03235","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1804.03235/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"1804.03235","created_at":"2026-07-05T01:28:39.924891+00:00"},{"alias_kind":"arxiv_version","alias_value":"1804.03235v2","created_at":"2026-07-05T01:28:39.924891+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1804.03235","created_at":"2026-07-05T01:28:39.924891+00:00"},{"alias_kind":"pith_short_12","alias_value":"SZ32ZXJZ5WEO","created_at":"2026-07-05T01:28:39.924891+00:00"},{"alias_kind":"pith_short_16","alias_value":"SZ32ZXJZ5WEO3OCJ","created_at":"2026-07-05T01:28:39.924891+00:00"},{"alias_kind":"pith_short_8","alias_value":"SZ32ZXJZ","created_at":"2026-07-05T01:28:39.924891+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24747","citing_title":"Scaling Laws for Task-Specific LLM Distillation","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2503.19786","citing_title":"Gemma 3 Technical Report","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21322","citing_title":"Optimized Federated Knowledge Distillation with Distributed Neural Architecture Search","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2507.06261","citing_title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21166","citing_title":"The Ratchet Effect in Silico: How Interaction Drives Cumulative Intelligence in Large Language Models","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2104.14294","citing_title":"Emerging Properties in Self-Supervised Vision Transformers","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2309.16588","citing_title":"Vision Transformers Need Registers","ref_index":182,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09356","citing_title":"Function-Space ADMM for Decentralized Federated Learning: A Control Theoretic Perspective","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23268","citing_title":"LatentBurst: A Fast and Efficient Multi Frame Super-Resolution for Hexadeca-Bayer Pattern CIS images","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05718","citing_title":"Enabling Federated Inference via Unsupervised Consensus Embedding","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2408.00118","citing_title":"Gemma 2: Improving Open Language Models at a Practical Size","ref_index":159,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21027","citing_title":"HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering","ref_index":178,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A","json":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A.json","graph_json":"https://pith.science/api/pith-number/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/graph.json","events_json":"https://pith.science/api/pith-number/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/events.json","paper":"https://pith.science/paper/SZ32ZXJZ"},"agent_actions":{"view_html":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A","download_json":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A.json","view_paper":"https://pith.science/paper/SZ32ZXJZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1804.03235&json=true","fetch_graph":"https://pith.science/api/pith-number/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/graph.json","fetch_events":"https://pith.science/api/pith-number/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/action/storage_attestation","attest_author":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/action/author_attestation","sign_citation":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/action/citation_signature","submit_replication":"https://pith.science/pith/SZ32ZXJZ5WEO3OCJNKVKHZAH4A/action/replication_record"}},"created_at":"2026-07-05T01:28:39.924891+00:00","updated_at":"2026-07-05T01:28:39.924891+00:00"}