def collect_params(all_params, params):
collected = utils.HParams()
for k in six.iterkeys(params.values()):
collected.add_hparam(k, getattr(all_params, k))
return collected
def default_params():
params = utils.HParams(
input="",
output="",
model="transformer",
vocab=["", ""],
pad="<pad>",
bos="<eos>",
eos="<eos>",
unk="<unk>",
# Dataset
batch_size=4096,
fixed_batch_size=False,
min_length=1,
max_length=256,
buffer_size=10000,
# Initialization
initializer_gain=1.0,
initializer="uniform_unit_scaling",
# Regularization
scale_l1=0.0,
scale_l2=0.0,
# Training
script="",
warmup_steps=4000,
train_steps=100000,
update_cycle=1,
optimizer="Adam",
adam_beta1=0.9,
adam_beta2=0.999,
adam_epsilon=1e-8,
adadelta_rho=0.95,
adadelta_epsilon=1e-6,
clipping="global_norm",
clip_grad_norm=5.0,
learning_rate=1.0,
learning_rate_schedule="linear_warmup_rsqrt_decay",
learning_rate_boundaries=[0],
learning_rate_values=[0.0],
device_list=[0],
embedding="",
# Validation
keep_top_k=50,
frequency=10,
# Checkpoint Saving
keep_checkpoint_max=20,
keep_top_checkpoint_max=5,
save_summary=True,
save_checkpoint_secs=0,
save_checkpoint_steps=1000,
)
return params
def default_params():
params = utils.HParams(
input=None,
output=None,
vocabulary=None,
embedding="",
# vocabulary specific
pad="<pad>",
bos="<bos>",
eos="<eos>",
unk="<unk>",
device=0,
decode_batch_size=128)
return params
def merge_params(params1, params2):
params = utils.HParams()
for (k, v) in six.iteritems(params1.values()):
params.add_hparam(k, v)
params_dict = params.values()
for (k, v) in six.iteritems(params2.values()):
if k in params_dict:
# Override
setattr(params, k, v)
else:
params.add_hparam(k, v)
return params
def base_params():
params = utils.HParams(pad="<pad>",
bos="<eos>",
eos="<eos>",
unk="<unk>",
feature_size=100,
hidden_size=200,
filter_size=800,
num_heads=8,
num_hidden_layers=10,
attention_dropout=0.0,
residual_dropout=0.1,
relu_dropout=0.0,
label_smoothing=0.1,
clip_grad_norm=0.0)
return params