{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:NXEW53IVRY3VIKYJHH2435DCKO","short_pith_number":"pith:NXEW53IV","schema_version":"1.0","canonical_sha256":"6dc96eed158e37542b0939f5cdf46253bf4aff246c4733dc3ab327b660447f17","source":{"kind":"arxiv","id":"2206.13236","version":1},"attestation_state":"computed","paper":{"title":"Pruned RNN-T for fast, memory-efficient ASR training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"eess.AS","authors_text":"Daniel Povey, Fangjun Kuang, Liyong Guo, Long Lin, Mingshuang Luo, Wei Kang, Zengwei Yao","submitted_at":"2022-06-23T12:18:03Z","abstract_excerpt":"The RNN-Transducer (RNN-T) framework for speech recognition has been growing in popularity, particularly for deployed real-time ASR systems, because it combines high accuracy with naturally streaming recognition. One of the drawbacks of RNN-T is that its loss function is relatively slow to compute, and can use a lot of memory. Excessive GPU memory usage can make it impractical to use RNN-T loss in cases where the vocabulary size is large: for example, for Chinese character-based ASR. We introduce a method for faster and more memory-efficient RNN-T loss computation. We first obtain pruning boun"},"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":"2206.13236","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"eess.AS","submitted_at":"2022-06-23T12:18:03Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"8e1209b519c4c1c647c50e48fd5aca6df1f4eb382898164171be0439ed87c5ff","abstract_canon_sha256":"d28267b37774b3e197a0f0b1f90799858dc8d7be1b343c3550f6c25ad75c2584"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:35:04.436000Z","signature_b64":"byIzF6Mhnf+Syu2TVqCVankx13TPiNpJ1kkkaF1Bh9NeRlGUZPG0B2FxasC6rTcJgx96PHZUmBcZo+Y7Q4JhBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6dc96eed158e37542b0939f5cdf46253bf4aff246c4733dc3ab327b660447f17","last_reissued_at":"2026-07-05T04:35:04.435564Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:35:04.435564Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pruned RNN-T for fast, memory-efficient ASR training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"eess.AS","authors_text":"Daniel Povey, Fangjun Kuang, Liyong Guo, Long Lin, Mingshuang Luo, Wei Kang, Zengwei Yao","submitted_at":"2022-06-23T12:18:03Z","abstract_excerpt":"The RNN-Transducer (RNN-T) framework for speech recognition has been growing in popularity, particularly for deployed real-time ASR systems, because it combines high accuracy with naturally streaming recognition. One of the drawbacks of RNN-T is that its loss function is relatively slow to compute, and can use a lot of memory. Excessive GPU memory usage can make it impractical to use RNN-T loss in cases where the vocabulary size is large: for example, for Chinese character-based ASR. We introduce a method for faster and more memory-efficient RNN-T loss computation. We first obtain pruning boun"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2206.13236","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/2206.13236/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":"2206.13236","created_at":"2026-07-05T04:35:04.435637+00:00"},{"alias_kind":"arxiv_version","alias_value":"2206.13236v1","created_at":"2026-07-05T04:35:04.435637+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2206.13236","created_at":"2026-07-05T04:35:04.435637+00:00"},{"alias_kind":"pith_short_12","alias_value":"NXEW53IVRY3V","created_at":"2026-07-05T04:35:04.435637+00:00"},{"alias_kind":"pith_short_16","alias_value":"NXEW53IVRY3VIKYJ","created_at":"2026-07-05T04:35:04.435637+00:00"},{"alias_kind":"pith_short_8","alias_value":"NXEW53IV","created_at":"2026-07-05T04:35:04.435637+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08486","citing_title":"TRADE: Transducer-Augmented Decoder for Speech LLM","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO","json":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO.json","graph_json":"https://pith.science/api/pith-number/NXEW53IVRY3VIKYJHH2435DCKO/graph.json","events_json":"https://pith.science/api/pith-number/NXEW53IVRY3VIKYJHH2435DCKO/events.json","paper":"https://pith.science/paper/NXEW53IV"},"agent_actions":{"view_html":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO","download_json":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO.json","view_paper":"https://pith.science/paper/NXEW53IV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2206.13236&json=true","fetch_graph":"https://pith.science/api/pith-number/NXEW53IVRY3VIKYJHH2435DCKO/graph.json","fetch_events":"https://pith.science/api/pith-number/NXEW53IVRY3VIKYJHH2435DCKO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO/action/storage_attestation","attest_author":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO/action/author_attestation","sign_citation":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO/action/citation_signature","submit_replication":"https://pith.science/pith/NXEW53IVRY3VIKYJHH2435DCKO/action/replication_record"}},"created_at":"2026-07-05T04:35:04.435637+00:00","updated_at":"2026-07-05T04:35:04.435637+00:00"}