ResearchBenchmarksStack Manipulation on Formal-language benchmark lengths 41-500 (test)Follow100AccuracySRNN54.2466.127889.88May 16, 2026Evaluation ResultsMethodMethodLinksAccuracySRNNArchitecture=Stack-RNNArchitecture=Stack-RNN2026.05100TRNNArchitecture=Tape-RNNArchitecture=Tape-RNN2026.05100LSTM2026.0559.1TransformerArchitecture=encoder-onlyArchitecture=encoder-only2026.0557.5PR-LSTM2026.0556.3RNN2026.0556