{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:J3232LA23VIC4TEUZIDMBAOJHR","short_pith_number":"pith:J3232LA2","schema_version":"1.0","canonical_sha256":"4ef5bd2c1add502e4c94ca06c081c93c5bd563108f483722a39860470b9862a0","source":{"kind":"arxiv","id":"2111.12763","version":1},"attestation_state":"computed","paper":{"title":"Sparse is Enough in Scaling Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Aakanksha Chowdhery, Afroz Mohiuddin, Henryk Michalewski, Jonni Kanerva, {\\L}ukasz Kaiser, Sebastian Jaszczur, Wojciech Gajewski","submitted_at":"2021-11-24T19:53:46Z","abstract_excerpt":"Large Transformer models yield impressive results on many tasks, but are expensive to train, or even fine-tune, and so slow at decoding that their use and study becomes out of reach. We address this problem by leveraging sparsity. We study sparse variants for all layers in the Transformer and propose Scaling Transformers, a family of next generation Transformer models that use sparse layers to scale efficiently and perform unbatched decoding much faster than the standard Transformer as we scale up the model size. Surprisingly, the sparse layers are enough to obtain the same perplexity as the s"},"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":"2111.12763","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-11-24T19:53:46Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"748c6759ba2187ea4960982bdd3616e726590fa34c44c7a3f8061de657137c47","abstract_canon_sha256":"3f19871e2148cd3f336f948197fb2147a50a440e2c59afc948b59422d70629c9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:34:58.002453Z","signature_b64":"DaT1lEP1cf0Qmuzcj0QwW35KOQWEphLB4M5xSdnWy6mIPPvqHmDLFiSor5eievHH3kVITiOtRbilriP/hfogBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4ef5bd2c1add502e4c94ca06c081c93c5bd563108f483722a39860470b9862a0","last_reissued_at":"2026-07-05T03:34:58.001951Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:34:58.001951Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sparse is Enough in Scaling Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Aakanksha Chowdhery, Afroz Mohiuddin, Henryk Michalewski, Jonni Kanerva, {\\L}ukasz Kaiser, Sebastian Jaszczur, Wojciech Gajewski","submitted_at":"2021-11-24T19:53:46Z","abstract_excerpt":"Large Transformer models yield impressive results on many tasks, but are expensive to train, or even fine-tune, and so slow at decoding that their use and study becomes out of reach. We address this problem by leveraging sparsity. We study sparse variants for all layers in the Transformer and propose Scaling Transformers, a family of next generation Transformer models that use sparse layers to scale efficiently and perform unbatched decoding much faster than the standard Transformer as we scale up the model size. Surprisingly, the sparse layers are enough to obtain the same perplexity as the s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2111.12763","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2111.12763/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":"2111.12763","created_at":"2026-07-05T03:34:58.002094+00:00"},{"alias_kind":"arxiv_version","alias_value":"2111.12763v1","created_at":"2026-07-05T03:34:58.002094+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2111.12763","created_at":"2026-07-05T03:34:58.002094+00:00"},{"alias_kind":"pith_short_12","alias_value":"J3232LA23VIC","created_at":"2026-07-05T03:34:58.002094+00:00"},{"alias_kind":"pith_short_16","alias_value":"J3232LA23VIC4TEU","created_at":"2026-07-05T03:34:58.002094+00:00"},{"alias_kind":"pith_short_8","alias_value":"J3232LA2","created_at":"2026-07-05T03:34:58.002094+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2502.05564","citing_title":"TabICL: A Tabular Foundation Model for In-Context Learning on Large Data","ref_index":250,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR","json":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR.json","graph_json":"https://pith.science/api/pith-number/J3232LA23VIC4TEUZIDMBAOJHR/graph.json","events_json":"https://pith.science/api/pith-number/J3232LA23VIC4TEUZIDMBAOJHR/events.json","paper":"https://pith.science/paper/J3232LA2"},"agent_actions":{"view_html":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR","download_json":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR.json","view_paper":"https://pith.science/paper/J3232LA2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2111.12763&json=true","fetch_graph":"https://pith.science/api/pith-number/J3232LA23VIC4TEUZIDMBAOJHR/graph.json","fetch_events":"https://pith.science/api/pith-number/J3232LA23VIC4TEUZIDMBAOJHR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR/action/storage_attestation","attest_author":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR/action/author_attestation","sign_citation":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR/action/citation_signature","submit_replication":"https://pith.science/pith/J3232LA23VIC4TEUZIDMBAOJHR/action/replication_record"}},"created_at":"2026-07-05T03:34:58.002094+00:00","updated_at":"2026-07-05T03:34:58.002094+00:00"}