Stephen Merity, Bryan McCann, Richard Socher
Motivation: Better baselines
Fast and open baselines are the
building blocks of future work
... yet rarely get any continued love.
Given the speed of progress within our field, and the many places progress arises, revisiting these baselines is important
Baseline numbers are frequently copy + pasted from other papers, rarely being re-run or improved
Revisiting RNN regularization
Standard activation regularization is easy,
even part of the Keras API ("activity_regularizer")
Activation regularization has nearly no impact on train time and is compatible with fast black box RNN implementations (i.e. NVIDIA cuDNN)
... yet they haven't been used recently?
Activation Regularization (AR)
Encourage small activations, penalizing any activations far from zero
For RNNs, simply add an additional loss,
where m is dropout mask and α is a scaler.
We found it's more effective when applied to the dropped output of the final RNN layer
\alpha L_2(m \cdot h_t)
Temporal Activation Regularization (TAR)
Loss penalizes changes between hidden states,
where h is the RNN output and β is a scaler.
Our work only penalizes the hidden output h of the RNN, leaving internal memory c unchanged.
\beta L_2(h_t - h_{t-1})
Penn Treebank Results
Note: these LSTMs have no recurrent regularization,
all other results here used variational recurrent dropout
WikiText-2 Results
Note: these LSTMs have no recurrent regularization,
all other results here used variational recurrent dropout
Related work
Independently LSTM LM baselines are revisited in Melis et al. (2017) and Merity et al. (2017)
Both, with negligible modifications to the LSTM, achieve strong gains over far more complex models on both Penn Treebank and WikiText-2
Melis et al. 2017 also investigate "hyper-parameter noise" / experimental setup of recent SotA models
Summary
Baselines should not be set in stone
As we improve the underlying fundamentals of our field, we should revisit them and improve them
Fast, open, and easy to extend baselines should be first class citizens for our field
We'll be releasing our PyTorch codebase for this and follow-up SotA (PTB/WT-2) numbers soon
Contact me online at @smerity / smerity.com
Check out
Revisiting Activation Regularization for Language RNNs
and our follow up work
Regularizing and Optimizing LSTM Language Models
Revisiting Activation Regularization for Language RNNs
By smerity
