{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:42XSFIWDNPIKL7ARZRH7V7HPVA","short_pith_number":"pith:42XSFIWD","schema_version":"1.0","canonical_sha256":"e6af22a2c36bd0a5fc11cc4ffafcefa825af2b8ec8ab3e1ba2122a9ac636e5fe","source":{"kind":"arxiv","id":"2106.07734","version":1},"attestation_state":"computed","paper":{"title":"CoDERT: Distilling Encoder Representations with Co-learning for Transducer-based Speech Recognition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","eess.AS"],"primary_cat":"cs.CL","authors_text":"Athanasios Mouchtaris, Brian King, Grant P. Strimel, Jasha Droppo, Rupak Vignesh Swaminathan","submitted_at":"2021-06-14T20:03:57Z","abstract_excerpt":"We propose a simple yet effective method to compress an RNN-Transducer (RNN-T) through the well-known knowledge distillation paradigm. We show that the transducer's encoder outputs naturally have a high entropy and contain rich information about acoustically similar word-piece confusions. This rich information is suppressed when combined with the lower entropy decoder outputs to produce the joint network logits. Consequently, we introduce an auxiliary loss to distill the encoder logits from a teacher transducer's encoder, and explore training strategies where this encoder distillation works ef"},"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":"2106.07734","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-06-14T20:03:57Z","cross_cats_sorted":["cs.LG","eess.AS"],"title_canon_sha256":"17b7fc94073ca20de5f8c1487973e4714ecd887218239fc9b6630b6f66872ec6","abstract_canon_sha256":"db1f9147c1d89ca52a11a1709c1a5717c8dc747a0644e48e61022bbad61a82d9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:49:33.301173Z","signature_b64":"ywbNxUVAWx940v/uktbaXmBL5vfL2WtRU5/0ybVxM4KBD2JcABTNveWS1l1wGi40zgZuS0JmPDu1mnGhNZGSCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e6af22a2c36bd0a5fc11cc4ffafcefa825af2b8ec8ab3e1ba2122a9ac636e5fe","last_reissued_at":"2026-07-05T02:49:33.300723Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:49:33.300723Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CoDERT: Distilling Encoder Representations with Co-learning for Transducer-based Speech Recognition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","eess.AS"],"primary_cat":"cs.CL","authors_text":"Athanasios Mouchtaris, Brian King, Grant P. Strimel, Jasha Droppo, Rupak Vignesh Swaminathan","submitted_at":"2021-06-14T20:03:57Z","abstract_excerpt":"We propose a simple yet effective method to compress an RNN-Transducer (RNN-T) through the well-known knowledge distillation paradigm. We show that the transducer's encoder outputs naturally have a high entropy and contain rich information about acoustically similar word-piece confusions. This rich information is suppressed when combined with the lower entropy decoder outputs to produce the joint network logits. Consequently, we introduce an auxiliary loss to distill the encoder logits from a teacher transducer's encoder, and explore training strategies where this encoder distillation works ef"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.07734","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/2106.07734/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":"2106.07734","created_at":"2026-07-05T02:49:33.300781+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.07734v1","created_at":"2026-07-05T02:49:33.300781+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.07734","created_at":"2026-07-05T02:49:33.300781+00:00"},{"alias_kind":"pith_short_12","alias_value":"42XSFIWDNPIK","created_at":"2026-07-05T02:49:33.300781+00:00"},{"alias_kind":"pith_short_16","alias_value":"42XSFIWDNPIKL7AR","created_at":"2026-07-05T02:49:33.300781+00:00"},{"alias_kind":"pith_short_8","alias_value":"42XSFIWD","created_at":"2026-07-05T02:49:33.300781+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/42XSFIWDNPIKL7ARZRH7V7HPVA","json":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA.json","graph_json":"https://pith.science/api/pith-number/42XSFIWDNPIKL7ARZRH7V7HPVA/graph.json","events_json":"https://pith.science/api/pith-number/42XSFIWDNPIKL7ARZRH7V7HPVA/events.json","paper":"https://pith.science/paper/42XSFIWD"},"agent_actions":{"view_html":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA","download_json":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA.json","view_paper":"https://pith.science/paper/42XSFIWD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.07734&json=true","fetch_graph":"https://pith.science/api/pith-number/42XSFIWDNPIKL7ARZRH7V7HPVA/graph.json","fetch_events":"https://pith.science/api/pith-number/42XSFIWDNPIKL7ARZRH7V7HPVA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA/action/storage_attestation","attest_author":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA/action/author_attestation","sign_citation":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA/action/citation_signature","submit_replication":"https://pith.science/pith/42XSFIWDNPIKL7ARZRH7V7HPVA/action/replication_record"}},"created_at":"2026-07-05T02:49:33.300781+00:00","updated_at":"2026-07-05T02:49:33.300781+00:00"}