{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:AZAD3IMYICHFHP24ITMTCBNRYF","short_pith_number":"pith:AZAD3IMY","schema_version":"1.0","canonical_sha256":"06403da198408e53bf5c44d93105b1c15a84590bcd717706fab7ae949dcad2ac","source":{"kind":"arxiv","id":"2210.09188","version":2},"attestation_state":"computed","paper":{"title":"Sub-8-bit quantization for on-device speech recognition: a regularization-free approach","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","eess.AS"],"primary_cat":"cs.SD","authors_text":"Athanasios Mouchtaris, Grant P. Strimel, Hieu Duy Nguyen, Kai Zhen, Martin Radfar, Nathan Susanj","submitted_at":"2022-10-17T15:42:26Z","abstract_excerpt":"For on-device automatic speech recognition (ASR), quantization aware training (QAT) is ubiquitous to achieve the trade-off between model predictive performance and efficiency. Among existing QAT methods, one major drawback is that the quantization centroids have to be predetermined and fixed. To overcome this limitation, we introduce a regularization-free, \"soft-to-hard\" compression mechanism with self-adjustable centroids in a mu-Law constrained space, resulting in a simpler yet more versatile quantization scheme, called General Quantizer (GQ). We apply GQ to ASR tasks using Recurrent Neural "},"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":"2210.09188","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SD","submitted_at":"2022-10-17T15:42:26Z","cross_cats_sorted":["cs.LG","eess.AS"],"title_canon_sha256":"15a216c2a253ee3ead2ec07ecafc6f5809f92675a789f59aa79c4e67f0dc6629","abstract_canon_sha256":"0a6e9b9dd944138a6b6fb93bf1b667602d39af4fe595f67249ca4783afedc173"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:12:16.645030Z","signature_b64":"viaMdY+OrVgvC6kNljjwuR//mtbLvPL6yirMbMiIqwCTmLw4qZgspBUlNkni7LOviMSjhYr/891VTQNA6e2kCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"06403da198408e53bf5c44d93105b1c15a84590bcd717706fab7ae949dcad2ac","last_reissued_at":"2026-07-05T05:12:16.644602Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:12:16.644602Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sub-8-bit quantization for on-device speech recognition: a regularization-free approach","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","eess.AS"],"primary_cat":"cs.SD","authors_text":"Athanasios Mouchtaris, Grant P. Strimel, Hieu Duy Nguyen, Kai Zhen, Martin Radfar, Nathan Susanj","submitted_at":"2022-10-17T15:42:26Z","abstract_excerpt":"For on-device automatic speech recognition (ASR), quantization aware training (QAT) is ubiquitous to achieve the trade-off between model predictive performance and efficiency. Among existing QAT methods, one major drawback is that the quantization centroids have to be predetermined and fixed. To overcome this limitation, we introduce a regularization-free, \"soft-to-hard\" compression mechanism with self-adjustable centroids in a mu-Law constrained space, resulting in a simpler yet more versatile quantization scheme, called General Quantizer (GQ). We apply GQ to ASR tasks using Recurrent Neural "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.09188","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/2210.09188/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":"2210.09188","created_at":"2026-07-05T05:12:16.644656+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.09188v2","created_at":"2026-07-05T05:12:16.644656+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.09188","created_at":"2026-07-05T05:12:16.644656+00:00"},{"alias_kind":"pith_short_12","alias_value":"AZAD3IMYICHF","created_at":"2026-07-05T05:12:16.644656+00:00"},{"alias_kind":"pith_short_16","alias_value":"AZAD3IMYICHFHP24","created_at":"2026-07-05T05:12:16.644656+00:00"},{"alias_kind":"pith_short_8","alias_value":"AZAD3IMY","created_at":"2026-07-05T05:12:16.644656+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/AZAD3IMYICHFHP24ITMTCBNRYF","json":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF.json","graph_json":"https://pith.science/api/pith-number/AZAD3IMYICHFHP24ITMTCBNRYF/graph.json","events_json":"https://pith.science/api/pith-number/AZAD3IMYICHFHP24ITMTCBNRYF/events.json","paper":"https://pith.science/paper/AZAD3IMY"},"agent_actions":{"view_html":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF","download_json":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF.json","view_paper":"https://pith.science/paper/AZAD3IMY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.09188&json=true","fetch_graph":"https://pith.science/api/pith-number/AZAD3IMYICHFHP24ITMTCBNRYF/graph.json","fetch_events":"https://pith.science/api/pith-number/AZAD3IMYICHFHP24ITMTCBNRYF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF/action/storage_attestation","attest_author":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF/action/author_attestation","sign_citation":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF/action/citation_signature","submit_replication":"https://pith.science/pith/AZAD3IMYICHFHP24ITMTCBNRYF/action/replication_record"}},"created_at":"2026-07-05T05:12:16.644656+00:00","updated_at":"2026-07-05T05:12:16.644656+00:00"}