Efficiently Modeling Long Sequences with Structured State Spaces

Benchmark Model Rank Results
language-modelling-on-wikitext-103S4#44Test perplexity: 21.28Number of params: 249M
sequential-image-classification-on-sequentialS4#4Permuted Accuracy: 98.70%Unpermuted Accuracy: 99.63%
sequential-image-classification-on-sequential-1S4#2Unpermuted Accuracy: 91.80%
speech-recognition-on-speech-commands-2S4#2Accuracy (%): 98.32