{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VJNSLZFZ55H5JRJRHQMU7S3UJE","short_pith_number":"pith:VJNSLZFZ","schema_version":"1.0","canonical_sha256":"aa5b25e4b9ef4fd4c5313c194fcb74490fd48a29531eee560e0d22234f0c498c","source":{"kind":"arxiv","id":"2405.21064","version":2},"attestation_state":"computed","paper":{"title":"Recurrent neural networks: vanishing and exploding gradients are not the end of the story","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Antonio Orvieto, Nicolas Zucchet","submitted_at":"2024-05-31T17:53:00Z","abstract_excerpt":"Recurrent neural networks (RNNs) notoriously struggle to learn long-term memories, primarily due to vanishing and exploding gradients. The recent success of state-space models (SSMs), a subclass of RNNs, to overcome such difficulties challenges our theoretical understanding. In this paper, we delve into the optimization challenges of RNNs and discover that, as the memory of a network increases, changes in its parameters result in increasingly large output variations, making gradient-based learning highly sensitive, even without exploding gradients. Our analysis further reveals the importance o"},"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":"2405.21064","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-31T17:53:00Z","cross_cats_sorted":["cs.AI","math.OC"],"title_canon_sha256":"9f8946088e551fbb88abf36ff9fc7dfb12166ba6dc8d356c738a86e9bea3418f","abstract_canon_sha256":"bfe280acd24619453be00f6ac154a121200f344bee5d54cd73583102a133908f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:31:03.492635Z","signature_b64":"v33/0WATYaYgGKnVmZ2kT0JlKnOzNmVmWShNrgWjCDLSr4YV7rii2uClFuoBLAMBCLL2hMqchtrd7eTkvTfZBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa5b25e4b9ef4fd4c5313c194fcb74490fd48a29531eee560e0d22234f0c498c","last_reissued_at":"2026-07-05T09:31:03.492136Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:31:03.492136Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Recurrent neural networks: vanishing and exploding gradients are not the end of the story","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Antonio Orvieto, Nicolas Zucchet","submitted_at":"2024-05-31T17:53:00Z","abstract_excerpt":"Recurrent neural networks (RNNs) notoriously struggle to learn long-term memories, primarily due to vanishing and exploding gradients. The recent success of state-space models (SSMs), a subclass of RNNs, to overcome such difficulties challenges our theoretical understanding. In this paper, we delve into the optimization challenges of RNNs and discover that, as the memory of a network increases, changes in its parameters result in increasingly large output variations, making gradient-based learning highly sensitive, even without exploding gradients. Our analysis further reveals the importance o"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.21064","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/2405.21064/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":"2405.21064","created_at":"2026-07-05T09:31:03.492196+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.21064v2","created_at":"2026-07-05T09:31:03.492196+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.21064","created_at":"2026-07-05T09:31:03.492196+00:00"},{"alias_kind":"pith_short_12","alias_value":"VJNSLZFZ55H5","created_at":"2026-07-05T09:31:03.492196+00:00"},{"alias_kind":"pith_short_16","alias_value":"VJNSLZFZ55H5JRJR","created_at":"2026-07-05T09:31:03.492196+00:00"},{"alias_kind":"pith_short_8","alias_value":"VJNSLZFZ","created_at":"2026-07-05T09:31:03.492196+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2506.06374","citing_title":"SiLIF: Structured State Space Model Dynamics and Parametrization for Spiking Neural Networks","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE","json":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE.json","graph_json":"https://pith.science/api/pith-number/VJNSLZFZ55H5JRJRHQMU7S3UJE/graph.json","events_json":"https://pith.science/api/pith-number/VJNSLZFZ55H5JRJRHQMU7S3UJE/events.json","paper":"https://pith.science/paper/VJNSLZFZ"},"agent_actions":{"view_html":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE","download_json":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE.json","view_paper":"https://pith.science/paper/VJNSLZFZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.21064&json=true","fetch_graph":"https://pith.science/api/pith-number/VJNSLZFZ55H5JRJRHQMU7S3UJE/graph.json","fetch_events":"https://pith.science/api/pith-number/VJNSLZFZ55H5JRJRHQMU7S3UJE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE/action/storage_attestation","attest_author":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE/action/author_attestation","sign_citation":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE/action/citation_signature","submit_replication":"https://pith.science/pith/VJNSLZFZ55H5JRJRHQMU7S3UJE/action/replication_record"}},"created_at":"2026-07-05T09:31:03.492196+00:00","updated_at":"2026-07-05T09:31:03.492196+00:00"}