{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:75IZLCJ3U2YIAJ5JKSI5UTUCPO","short_pith_number":"pith:75IZLCJ3","schema_version":"1.0","canonical_sha256":"ff5195893ba6b08027a95491da4e827b8563429cbbdd6d11eeab331f85f04b86","source":{"kind":"arxiv","id":"2010.11524","version":5},"attestation_state":"computed","paper":{"title":"SlimIPL: Language-Model-Free Iterative Pseudo-Labeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Gabriel Synnaeve, Jacob Kahn, Qiantong Xu, Ronan Collobert, Tatiana Likhomanenko","submitted_at":"2020-10-22T08:36:33Z","abstract_excerpt":"Recent results in end-to-end automatic speech recognition have demonstrated the efficacy of pseudo-labeling for semi-supervised models trained both with Connectionist Temporal Classification (CTC) and Sequence-to-Sequence (seq2seq) losses. Iterative Pseudo-Labeling (IPL), which continuously trains a single model using pseudo-labels iteratively re-generated as the model learns, has been shown to further improve performance in ASR. We improve upon the IPL algorithm: as the model learns, we propose to iteratively re-generate transcriptions with hard labels (the most probable tokens), that is, wit"},"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":"2010.11524","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-10-22T08:36:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"8a548dff850abe553794496c00e85b93c8395c57b2d1f168f43772b61a93717d","abstract_canon_sha256":"bddddb738adac3e26e6e4701b0123968a856803485fdd625085c7990ee79697d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:09:43.430100Z","signature_b64":"6+mH2DamvmCco6uKBZfS2BYgoVXIrXTCjgULbYn/I8ruttCpLOicfMLosQND26MrGy13FGxgSHbKA7cZHbITAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff5195893ba6b08027a95491da4e827b8563429cbbdd6d11eeab331f85f04b86","last_reissued_at":"2026-07-05T03:09:43.429572Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:09:43.429572Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SlimIPL: Language-Model-Free Iterative Pseudo-Labeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Gabriel Synnaeve, Jacob Kahn, Qiantong Xu, Ronan Collobert, Tatiana Likhomanenko","submitted_at":"2020-10-22T08:36:33Z","abstract_excerpt":"Recent results in end-to-end automatic speech recognition have demonstrated the efficacy of pseudo-labeling for semi-supervised models trained both with Connectionist Temporal Classification (CTC) and Sequence-to-Sequence (seq2seq) losses. Iterative Pseudo-Labeling (IPL), which continuously trains a single model using pseudo-labels iteratively re-generated as the model learns, has been shown to further improve performance in ASR. We improve upon the IPL algorithm: as the model learns, we propose to iteratively re-generate transcriptions with hard labels (the most probable tokens), that is, wit"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2010.11524","kind":"arxiv","version":5},"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/2010.11524/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":"2010.11524","created_at":"2026-07-05T03:09:43.429628+00:00"},{"alias_kind":"arxiv_version","alias_value":"2010.11524v5","created_at":"2026-07-05T03:09:43.429628+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2010.11524","created_at":"2026-07-05T03:09:43.429628+00:00"},{"alias_kind":"pith_short_12","alias_value":"75IZLCJ3U2YI","created_at":"2026-07-05T03:09:43.429628+00:00"},{"alias_kind":"pith_short_16","alias_value":"75IZLCJ3U2YIAJ5J","created_at":"2026-07-05T03:09:43.429628+00:00"},{"alias_kind":"pith_short_8","alias_value":"75IZLCJ3","created_at":"2026-07-05T03:09:43.429628+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.04814","citing_title":"Pitch Accent Detection improves Pretrained Automatic Speech Recognition","ref_index":52,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO","json":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO.json","graph_json":"https://pith.science/api/pith-number/75IZLCJ3U2YIAJ5JKSI5UTUCPO/graph.json","events_json":"https://pith.science/api/pith-number/75IZLCJ3U2YIAJ5JKSI5UTUCPO/events.json","paper":"https://pith.science/paper/75IZLCJ3"},"agent_actions":{"view_html":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO","download_json":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO.json","view_paper":"https://pith.science/paper/75IZLCJ3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2010.11524&json=true","fetch_graph":"https://pith.science/api/pith-number/75IZLCJ3U2YIAJ5JKSI5UTUCPO/graph.json","fetch_events":"https://pith.science/api/pith-number/75IZLCJ3U2YIAJ5JKSI5UTUCPO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO/action/storage_attestation","attest_author":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO/action/author_attestation","sign_citation":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO/action/citation_signature","submit_replication":"https://pith.science/pith/75IZLCJ3U2YIAJ5JKSI5UTUCPO/action/replication_record"}},"created_at":"2026-07-05T03:09:43.429628+00:00","updated_at":"2026-07-05T03:09:43.429628+00:00"}