{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:QSVSRF2UMCXIRW2HLPEYXEQKJP","short_pith_number":"pith:QSVSRF2U","schema_version":"1.0","canonical_sha256":"84ab28975460ae88db475bc98b920a4bd07f5a8693de0103433dfb2016e41a70","source":{"kind":"arxiv","id":"2102.11417","version":2},"attestation_state":"computed","paper":{"title":"Parallelizing Legendre Memory Unit Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chris Eliasmith, Narsimha Chilkuri","submitted_at":"2021-02-22T23:43:47Z","abstract_excerpt":"Recently, a new recurrent neural network (RNN) named the Legendre Memory Unit (LMU) was proposed and shown to achieve state-of-the-art performance on several benchmark datasets. Here we leverage the linear time-invariant (LTI) memory component of the LMU to construct a simplified variant that can be parallelized during training (and yet executed as an RNN during inference), thus overcoming a well known limitation of training RNNs on GPUs. We show that this reformulation that aids parallelizing, which can be applied generally to any deep network whose recurrent components are linear, makes trai"},"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":"2102.11417","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-22T23:43:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"664d635590e8108970db58a3f6b72e92e9b7cfaf197769f52fe199f948a75b86","abstract_canon_sha256":"4d5055561f630e84bdd2bd98bd066e9c047763be394d0d5f3143d035840174b2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:39:14.972015Z","signature_b64":"CgH7PoSFx28n0ad3w5J5PzS9t3Fme4/Wc63mdw5kxL24xb91V3Gxt3HXhBiP64qhmltw7YQ4FaxCK3iiNWPzAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"84ab28975460ae88db475bc98b920a4bd07f5a8693de0103433dfb2016e41a70","last_reissued_at":"2026-07-05T02:39:14.971535Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:39:14.971535Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Parallelizing Legendre Memory Unit Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chris Eliasmith, Narsimha Chilkuri","submitted_at":"2021-02-22T23:43:47Z","abstract_excerpt":"Recently, a new recurrent neural network (RNN) named the Legendre Memory Unit (LMU) was proposed and shown to achieve state-of-the-art performance on several benchmark datasets. Here we leverage the linear time-invariant (LTI) memory component of the LMU to construct a simplified variant that can be parallelized during training (and yet executed as an RNN during inference), thus overcoming a well known limitation of training RNNs on GPUs. We show that this reformulation that aids parallelizing, which can be applied generally to any deep network whose recurrent components are linear, makes trai"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.11417","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/2102.11417/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":"2102.11417","created_at":"2026-07-05T02:39:14.971599+00:00"},{"alias_kind":"arxiv_version","alias_value":"2102.11417v2","created_at":"2026-07-05T02:39:14.971599+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.11417","created_at":"2026-07-05T02:39:14.971599+00:00"},{"alias_kind":"pith_short_12","alias_value":"QSVSRF2UMCXI","created_at":"2026-07-05T02:39:14.971599+00:00"},{"alias_kind":"pith_short_16","alias_value":"QSVSRF2UMCXIRW2H","created_at":"2026-07-05T02:39:14.971599+00:00"},{"alias_kind":"pith_short_8","alias_value":"QSVSRF2U","created_at":"2026-07-05T02:39:14.971599+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.09827","citing_title":"The Good, The Efficient and the Inductive Biases: Exploring Efficiency in Deep Learning Through the Use of Inductive Biases","ref_index":55,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP","json":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP.json","graph_json":"https://pith.science/api/pith-number/QSVSRF2UMCXIRW2HLPEYXEQKJP/graph.json","events_json":"https://pith.science/api/pith-number/QSVSRF2UMCXIRW2HLPEYXEQKJP/events.json","paper":"https://pith.science/paper/QSVSRF2U"},"agent_actions":{"view_html":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP","download_json":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP.json","view_paper":"https://pith.science/paper/QSVSRF2U","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2102.11417&json=true","fetch_graph":"https://pith.science/api/pith-number/QSVSRF2UMCXIRW2HLPEYXEQKJP/graph.json","fetch_events":"https://pith.science/api/pith-number/QSVSRF2UMCXIRW2HLPEYXEQKJP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP/action/storage_attestation","attest_author":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP/action/author_attestation","sign_citation":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP/action/citation_signature","submit_replication":"https://pith.science/pith/QSVSRF2UMCXIRW2HLPEYXEQKJP/action/replication_record"}},"created_at":"2026-07-05T02:39:14.971599+00:00","updated_at":"2026-07-05T02:39:14.971599+00:00"}