{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FTDHCNMOACEPBRWEMPNKC6UU6G","short_pith_number":"pith:FTDHCNMO","schema_version":"1.0","canonical_sha256":"2cc671358e0088f0c6c463daa17a94f1bb20b9ff2b360fe762438c2116d5fc5e","source":{"kind":"arxiv","id":"2412.07752","version":3},"attestation_state":"computed","paper":{"title":"FlashRNN: I/O-Aware Optimization of Traditional RNNs on modern hardware","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Korbinian P\\\"oppel, Maximilian Beck, Sepp Hochreiter","submitted_at":"2024-12-10T18:50:37Z","abstract_excerpt":"While Transformers and other sequence-parallelizable neural network architectures seem like the current state of the art in sequence modeling, they specifically lack state-tracking capabilities. These are important for time-series tasks and logical reasoning. Traditional RNNs like LSTMs and GRUs, as well as modern variants like sLSTM do have these capabilities at the cost of strictly sequential processing. While this is often seen as a strong limitation, we show how fast these networks can get with our hardware-optimization FlashRNN in Triton and CUDA, optimizing kernels to the register level "},"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":"2412.07752","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-12-10T18:50:37Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3725cab91cefe5c85aa30927105dd652250098d077ac1659d532581eac5a57be","abstract_canon_sha256":"14eeb56c7a4db80792637f41fb25bf21e5422ea57aea6ccc3cfc7398a20dedac"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:30:28.019193Z","signature_b64":"8mMvGqKaJzF/WKP+Mi9dNl+7E5c5QOZ3gO2lkdMken8FhfeJnvZVEtJ2IJuxgE5+6XQ+OCNkDd7o7r1Pioh0DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2cc671358e0088f0c6c463daa17a94f1bb20b9ff2b360fe762438c2116d5fc5e","last_reissued_at":"2026-07-05T10:30:28.018687Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:30:28.018687Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FlashRNN: I/O-Aware Optimization of Traditional RNNs on modern hardware","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Korbinian P\\\"oppel, Maximilian Beck, Sepp Hochreiter","submitted_at":"2024-12-10T18:50:37Z","abstract_excerpt":"While Transformers and other sequence-parallelizable neural network architectures seem like the current state of the art in sequence modeling, they specifically lack state-tracking capabilities. These are important for time-series tasks and logical reasoning. Traditional RNNs like LSTMs and GRUs, as well as modern variants like sLSTM do have these capabilities at the cost of strictly sequential processing. While this is often seen as a strong limitation, we show how fast these networks can get with our hardware-optimization FlashRNN in Triton and CUDA, optimizing kernels to the register level "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.07752","kind":"arxiv","version":3},"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/2412.07752/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":"2412.07752","created_at":"2026-07-05T10:30:28.018751+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.07752v3","created_at":"2026-07-05T10:30:28.018751+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.07752","created_at":"2026-07-05T10:30:28.018751+00:00"},{"alias_kind":"pith_short_12","alias_value":"FTDHCNMOACEP","created_at":"2026-07-05T10:30:28.018751+00:00"},{"alias_kind":"pith_short_16","alias_value":"FTDHCNMOACEPBRWE","created_at":"2026-07-05T10:30:28.018751+00:00"},{"alias_kind":"pith_short_8","alias_value":"FTDHCNMO","created_at":"2026-07-05T10:30:28.018751+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20411","citing_title":"Direct Advantage Estimation for Scalable and Sample-efficient Deep Reinforcement Learning","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2603.14360","citing_title":"M$^2$RNN: Non-Linear RNNs with Matrix-Valued States for Scalable Language Modeling","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G","json":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G.json","graph_json":"https://pith.science/api/pith-number/FTDHCNMOACEPBRWEMPNKC6UU6G/graph.json","events_json":"https://pith.science/api/pith-number/FTDHCNMOACEPBRWEMPNKC6UU6G/events.json","paper":"https://pith.science/paper/FTDHCNMO"},"agent_actions":{"view_html":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G","download_json":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G.json","view_paper":"https://pith.science/paper/FTDHCNMO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.07752&json=true","fetch_graph":"https://pith.science/api/pith-number/FTDHCNMOACEPBRWEMPNKC6UU6G/graph.json","fetch_events":"https://pith.science/api/pith-number/FTDHCNMOACEPBRWEMPNKC6UU6G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G/action/storage_attestation","attest_author":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G/action/author_attestation","sign_citation":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G/action/citation_signature","submit_replication":"https://pith.science/pith/FTDHCNMOACEPBRWEMPNKC6UU6G/action/replication_record"}},"created_at":"2026-07-05T10:30:28.018751+00:00","updated_at":"2026-07-05T10:30:28.018751+00:00"}