{"paper":{"title":"An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A simple convolutional architecture outperforms LSTMs on diverse sequence tasks while showing longer effective memory.","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"J. Zico Kolter, Shaojie Bai, Vladlen Koltun","submitted_at":"2018-03-04T00:20:29Z","abstract_excerpt":"For most deep learning practitioners, sequence modeling is synonymous with recurrent networks. Yet recent results indicate that convolutional architectures can outperform recurrent networks on tasks such as audio synthesis and machine translation. Given a new sequence modeling task or dataset, which architecture should one use? We conduct a systematic evaluation of generic convolutional and recurrent architectures for sequence modeling. The models are evaluated across a broad range of standard tasks that are commonly used to benchmark recurrent networks. Our results indicate that a simple conv"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our results indicate that a simple convolutional architecture outperforms canonical recurrent networks such as LSTMs across a diverse range of tasks and datasets, while demonstrating longer effective memory.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the chosen tasks and datasets are representative of general sequence modeling challenges and that the generic convolutional and recurrent architectures are implemented and compared fairly without hidden advantages.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A simple convolutional network outperforms LSTMs across diverse sequence tasks while showing longer effective memory.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A simple convolutional architecture outperforms LSTMs on diverse sequence tasks while showing longer effective memory.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"a7893b8bc1fc49f86d22a6cf05b90ebac13add52188df11b8a7ef149f4b40cf6"},"source":{"id":"1803.01271","kind":"arxiv","version":2},"verdict":{"id":"0866af40-487f-48c5-ae7b-68eab1b2d011","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T19:30:04.145631Z","strongest_claim":"Our results indicate that a simple convolutional architecture outperforms canonical recurrent networks such as LSTMs across a diverse range of tasks and datasets, while demonstrating longer effective memory.","one_line_summary":"A simple convolutional network outperforms LSTMs across diverse sequence tasks while showing longer effective memory.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the chosen tasks and datasets are representative of general sequence modeling challenges and that the generic convolutional and recurrent architectures are implemented and compared fairly without hidden advantages.","pith_extraction_headline":"A simple convolutional architecture outperforms LSTMs on diverse sequence tasks while showing longer effective memory."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1803.01271/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":1,"snapshot_sha256":"bc59c0cce00c67ebdb6fa9de07055e0c5c77d51c7c2633f1f5a6e1977e3b54cf"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}