Recurrent Neural Network Regularization
2014-09-08Code Available0· sign in to hype
Wojciech Zaremba, Ilya Sutskever, Oriol Vinyals
Code Available — Be the first to reproduce this paper.
ReproduceCode
- github.com/wojzaremba/lstmOfficialIn papernone★ 0
- github.com/jincan333/lotpytorch★ 33
- github.com/rgarzonj/LSTMstf★ 0
- github.com/Goodideax/lstm-negtivepytorch★ 0
- github.com/martin-gorner/tensorflow-rnn-shakespearetf★ 0
- github.com/shivam13juna/Sequence_Prediction_LSTM_CHARtf★ 0
- github.com/hjc18/language_modeling_lstmpytorch★ 0
- github.com/MindSpore-scientific/code-5/tree/main/ReSegmindspore★ 0
- github.com/nbansal90/bAbi_QAnone★ 0
- github.com/MindSpore-scientific/code-14/tree/main/ReSegmindspore★ 0
Abstract
We present a simple regularization technique for Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM) units. Dropout, the most successful technique for regularizing neural networks, does not work well with RNNs and LSTMs. In this paper, we show how to correctly apply dropout to LSTMs, and show that it substantially reduces overfitting on a variety of tasks. These tasks include language modeling, speech recognition, image caption generation, and machine translation.
Tasks
Benchmark Results
| Dataset | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| Penn Treebank (Word Level) | Zaremba et al. (2014) - LSTM (large) | Test perplexity | 78.4 | — | Unverified |
| Penn Treebank (Word Level) | Zaremba et al. (2014) - LSTM (medium) | Test perplexity | 82.7 | — | Unverified |