{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:SR5ZP65LC3QQM76XBGAIBPAB5E","short_pith_number":"pith:SR5ZP65L","schema_version":"1.0","canonical_sha256":"947b97fbab16e1067fd7098080bc01e92994a6cb06e8f52fccc82751a07c0703","source":{"kind":"arxiv","id":"2009.07453","version":2},"attestation_state":"computed","paper":{"title":"Extremely Low Bit Transformer Quantization for On-Device Neural Machine Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Baeseong Park, Byeongwook Kim, Dongsoo Lee, Insoo Chung, Sangha Kim, Se Jung Kwon, Yongkweon Jeon, Yoonjung Choi","submitted_at":"2020-09-16T03:58:01Z","abstract_excerpt":"The deployment of widely used Transformer architecture is challenging because of heavy computation load and memory overhead during inference, especially when the target device is limited in computational resources such as mobile or edge devices. Quantization is an effective technique to address such challenges. Our analysis shows that for a given number of quantization bits, each block of Transformer contributes to translation quality and inference computations in different manners. Moreover, even inside an embedding block, each word presents vastly different contributions. Correspondingly, we"},"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":"2009.07453","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-09-16T03:58:01Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"9c6be7eae1cae91086badddd2cfdeaf9634d4bdef24b1761257a0866a255cc28","abstract_canon_sha256":"1f22ff2374577f46d191dba6e7b325b5026c362a3ea03dd75195ec733d2c5924"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:42:21.940036Z","signature_b64":"z1LfQ8GPasltpYy2LWzVh+a5UkFLdYXNmylqgonbQ4yev+7QOEIbstnbWMamAm9ljoh5THL2hpWnQwhGvBQZAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"947b97fbab16e1067fd7098080bc01e92994a6cb06e8f52fccc82751a07c0703","last_reissued_at":"2026-07-05T01:42:21.939582Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:42:21.939582Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Extremely Low Bit Transformer Quantization for On-Device Neural Machine Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Baeseong Park, Byeongwook Kim, Dongsoo Lee, Insoo Chung, Sangha Kim, Se Jung Kwon, Yongkweon Jeon, Yoonjung Choi","submitted_at":"2020-09-16T03:58:01Z","abstract_excerpt":"The deployment of widely used Transformer architecture is challenging because of heavy computation load and memory overhead during inference, especially when the target device is limited in computational resources such as mobile or edge devices. Quantization is an effective technique to address such challenges. Our analysis shows that for a given number of quantization bits, each block of Transformer contributes to translation quality and inference computations in different manners. Moreover, even inside an embedding block, each word presents vastly different contributions. Correspondingly, we"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2009.07453","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/2009.07453/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":"2009.07453","created_at":"2026-07-05T01:42:21.939646+00:00"},{"alias_kind":"arxiv_version","alias_value":"2009.07453v2","created_at":"2026-07-05T01:42:21.939646+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2009.07453","created_at":"2026-07-05T01:42:21.939646+00:00"},{"alias_kind":"pith_short_12","alias_value":"SR5ZP65LC3QQ","created_at":"2026-07-05T01:42:21.939646+00:00"},{"alias_kind":"pith_short_16","alias_value":"SR5ZP65LC3QQM76X","created_at":"2026-07-05T01:42:21.939646+00:00"},{"alias_kind":"pith_short_8","alias_value":"SR5ZP65L","created_at":"2026-07-05T01:42:21.939646+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/SR5ZP65LC3QQM76XBGAIBPAB5E","json":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E.json","graph_json":"https://pith.science/api/pith-number/SR5ZP65LC3QQM76XBGAIBPAB5E/graph.json","events_json":"https://pith.science/api/pith-number/SR5ZP65LC3QQM76XBGAIBPAB5E/events.json","paper":"https://pith.science/paper/SR5ZP65L"},"agent_actions":{"view_html":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E","download_json":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E.json","view_paper":"https://pith.science/paper/SR5ZP65L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2009.07453&json=true","fetch_graph":"https://pith.science/api/pith-number/SR5ZP65LC3QQM76XBGAIBPAB5E/graph.json","fetch_events":"https://pith.science/api/pith-number/SR5ZP65LC3QQM76XBGAIBPAB5E/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E/action/storage_attestation","attest_author":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E/action/author_attestation","sign_citation":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E/action/citation_signature","submit_replication":"https://pith.science/pith/SR5ZP65LC3QQM76XBGAIBPAB5E/action/replication_record"}},"created_at":"2026-07-05T01:42:21.939646+00:00","updated_at":"2026-07-05T01:42:21.939646+00:00"}