{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:6X74ADG45X5CYCJRGHTXA555QL","short_pith_number":"pith:6X74ADG4","schema_version":"1.0","canonical_sha256":"f5ffc00cdcedfa2c093131e77077bd82ca57de27f307bb87bc72ca486c29a770","source":{"kind":"arxiv","id":"2106.04718","version":2},"attestation_state":"computed","paper":{"title":"FastSeq: Make Sequence Generation Faster","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Bingyu Chi, Desheng Cui, Fei Hu, Jiusheng Chen, Nan Duan, Nikhil Bhendawade, Ruofei Zhang, Ting Ye, Yeyun Gong, Yu Yan","submitted_at":"2021-06-08T22:25:28Z","abstract_excerpt":"Transformer-based models have made tremendous impacts in natural language generation. However the inference speed is a bottleneck due to large model size and intensive computing involved in auto-regressive decoding process. We develop FastSeq framework to accelerate sequence generation without accuracy loss. The proposed optimization techniques include an attention cache optimization, an efficient algorithm for detecting repeated n-grams, and an asynchronous generation pipeline with parallel I/O. These optimizations are general enough to be applicable to Transformer-based models (e.g., T5, GPT"},"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":"2106.04718","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-06-08T22:25:28Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"bb557af0c7dc3a52ffbec77aba45ccf09746aa2d1686e1114af3341eaa697c6e","abstract_canon_sha256":"befc9121d650ebd8c1782098c18774368eb6b223e161737a8aa2c06570cca7f0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:57:23.044697Z","signature_b64":"aL/3TvqDezu4IvPL3xc9UNtQSp0nlR2WrOtaHbZopWZ5Q/ssSkrp29cCKdEauhWoXexjYsNDuLy79vonfTdoAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f5ffc00cdcedfa2c093131e77077bd82ca57de27f307bb87bc72ca486c29a770","last_reissued_at":"2026-07-05T02:57:23.044231Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:57:23.044231Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FastSeq: Make Sequence Generation Faster","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Bingyu Chi, Desheng Cui, Fei Hu, Jiusheng Chen, Nan Duan, Nikhil Bhendawade, Ruofei Zhang, Ting Ye, Yeyun Gong, Yu Yan","submitted_at":"2021-06-08T22:25:28Z","abstract_excerpt":"Transformer-based models have made tremendous impacts in natural language generation. However the inference speed is a bottleneck due to large model size and intensive computing involved in auto-regressive decoding process. We develop FastSeq framework to accelerate sequence generation without accuracy loss. The proposed optimization techniques include an attention cache optimization, an efficient algorithm for detecting repeated n-grams, and an asynchronous generation pipeline with parallel I/O. These optimizations are general enough to be applicable to Transformer-based models (e.g., T5, GPT"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.04718","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/2106.04718/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":"2106.04718","created_at":"2026-07-05T02:57:23.044288+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.04718v2","created_at":"2026-07-05T02:57:23.044288+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.04718","created_at":"2026-07-05T02:57:23.044288+00:00"},{"alias_kind":"pith_short_12","alias_value":"6X74ADG45X5C","created_at":"2026-07-05T02:57:23.044288+00:00"},{"alias_kind":"pith_short_16","alias_value":"6X74ADG45X5CYCJR","created_at":"2026-07-05T02:57:23.044288+00:00"},{"alias_kind":"pith_short_8","alias_value":"6X74ADG4","created_at":"2026-07-05T02:57:23.044288+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/6X74ADG45X5CYCJRGHTXA555QL","json":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL.json","graph_json":"https://pith.science/api/pith-number/6X74ADG45X5CYCJRGHTXA555QL/graph.json","events_json":"https://pith.science/api/pith-number/6X74ADG45X5CYCJRGHTXA555QL/events.json","paper":"https://pith.science/paper/6X74ADG4"},"agent_actions":{"view_html":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL","download_json":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL.json","view_paper":"https://pith.science/paper/6X74ADG4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.04718&json=true","fetch_graph":"https://pith.science/api/pith-number/6X74ADG45X5CYCJRGHTXA555QL/graph.json","fetch_events":"https://pith.science/api/pith-number/6X74ADG45X5CYCJRGHTXA555QL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL/action/storage_attestation","attest_author":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL/action/author_attestation","sign_citation":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL/action/citation_signature","submit_replication":"https://pith.science/pith/6X74ADG45X5CYCJRGHTXA555QL/action/replication_record"}},"created_at":"2026-07-05T02:57:23.044288+00:00","updated_at":"2026-07-05T02:57:23.044288+00:00"}