commit 1c5e3be221e5d8321a68398cdcebce30335e3489
parent e8a05d2a1d0c4481fc527d55041a95c452cfc94a
Author: umhau <umhau@users.noreply.github.com>
Date: Sat, 18 Feb 2017 18:58:43 -0500
adjust default parameters
Diffstat:
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/sample.lua b/sample.lua
@@ -37,7 +37,7 @@ cmd:option('-opencl',0,'use OpenCL (instead of CUDA)')
cmd:option('-verbose',1,'set to 0 to ONLY print the sampled text, no diagnostics')
cmd:option('-skip_unk',0,'whether to skip UNK tokens when sampling')
cmd:option('-input_loop',0,'whether to read new seed text from stdin after having finished sampling')
-cmd:option('-word_level',0,'whether to operate on the word level, instead of character level (0: use chars, 1: use words)') --todo: set this in checkpoint
+cmd:option('-word_level',1,'whether to operate on the word level, instead of character level (0: use chars, 1: use words)') --todo: set this in checkpoint
cmd:text()
-- parse input params
diff --git a/train.lua b/train.lua
@@ -46,11 +46,11 @@ cmd:option('-learning_rate',2e-3,'learning rate')
cmd:option('-learning_rate_decay',0.97,'learning rate decay')
cmd:option('-learning_rate_decay_after',10,'in number of epochs, when to start decaying the learning rate')
cmd:option('-decay_rate',0.95,'decay rate for rmsprop')
-cmd:option('-dropout',0,'dropout for regularization, used after each RNN hidden layer. 0 = no dropout')
+cmd:option('-dropout',0,'dropout for regularization, used after each RNN hidden layer. 0 = no dropout, .3 = 30% dropout')
cmd:option('-recurrent_dropout',0,'dropout for regularization, used on recurrent connections. 0 = no dropout')
cmd:option('-seq_length',50,'number of timesteps to unroll for')
cmd:option('-batch_size',50,'number of sequences to train on in parallel')
-cmd:option('-max_epochs',0.01,'number of full passes through the training data')
+cmd:option('-max_epochs',50,'number of full passes through the training data')
cmd:option('-grad_clip',5,'clip gradients at this value')
cmd:option('-train_frac',0.95,'fraction of data that goes into train set')
cmd:option('-val_frac',0.05,'fraction of data that goes into validation set')
@@ -66,7 +66,7 @@ cmd:option('-accurate_gpu_timing',0,'set this flag to 1 to get precise timings w
-- GPU/CPU
cmd:option('-gpuid',-1,'which gpu to use. -1 = use CPU')
cmd:option('-opencl',0,'use OpenCL (instead of CUDA)')
-cmd:option('-word_level',0,'whether to operate on the word level, instead of character level (0: use chars, 1: use words)')
+cmd:option('-word_level',1,'whether to operate on the word level, instead of character level (0: use chars, 1: use words)')
cmd:option('-threshold',0,'minimum number of occurences a token must have to be included (ignored if -word_level is 0)')
cmd:option('-glove',0,'whether or not to use GloVe embeddings')
cmd:option('-optimizer','eamsgd','which optimizer to use: adam or rmsprop')