diff options
author | Thomas Mesnard <thomas.mesnard@ens.fr> | 2016-03-01 00:27:15 +0100 |
---|---|---|
committer | Thomas Mesnard <thomas.mesnard@ens.fr> | 2016-03-02 09:28:39 +0100 |
commit | f31caf61be87850f3afcd367d6eb9521b2f613da (patch) | |
tree | 2bcceeb702ef0d35bfdc925977797c40290b6966 /config/deep_bidir_lstm_2x128.py | |
download | deepmind-qa-f31caf61be87850f3afcd367d6eb9521b2f613da.tar.gz deepmind-qa-f31caf61be87850f3afcd367d6eb9521b2f613da.zip |
Initial commit
Diffstat (limited to 'config/deep_bidir_lstm_2x128.py')
-rw-r--r-- | config/deep_bidir_lstm_2x128.py | 37 |
1 files changed, 37 insertions, 0 deletions
diff --git a/config/deep_bidir_lstm_2x128.py b/config/deep_bidir_lstm_2x128.py new file mode 100644 index 0000000..f07f43f --- /dev/null +++ b/config/deep_bidir_lstm_2x128.py @@ -0,0 +1,37 @@ +from blocks.algorithms import BasicMomentum, AdaDelta, RMSProp, Adam, CompositeRule, StepClipping +from blocks.initialization import IsotropicGaussian, Constant +from blocks.bricks import Tanh + +from model.deep_bidir_lstm import Model + + +batch_size = 32 +sort_batch_count = 20 + +shuffle_questions = True +shuffle_entities = True + +concat_ctx_and_question = True +concat_question_before = True ## should not matter for bidirectionnal network + +embed_size = 200 + +lstm_size = [128, 128] +skip_connections = True + +n_entities = 550 +out_mlp_hidden = [] +out_mlp_activations = [] + +step_rule = CompositeRule([RMSProp(decay_rate=0.95, learning_rate=5e-5), + BasicMomentum(momentum=0.9)]) + +dropout = 0.1 +w_noise = 0.05 + +valid_freq = 1000 +save_freq = 1000 +print_freq = 100 + +weights_init = IsotropicGaussian(0.01) +biases_init = Constant(0.) |