def test_transformer_ensemble_inference(test_dir):
from src.bin import ensemble_translate
from src.utils.common_utils import GlobalNames
config_path = "./unittests/configs/test_odc_loss.yaml"
saveto = os.path.join(test_dir, "save")
model_name = test_utils.get_model_name(config_path)
model_path = os.path.join(
saveto, model_name + GlobalNames.MY_BEST_MODEL_SUFFIX + ".final")
model_path = [model_path for _ in range(3)]
source_path = "./unittests/data/dev/zh.0"
batch_size = 3
beam_size = 3
ensemble_translate.run(model_name=model_name,
source_path=source_path,
batch_size=batch_size,
beam_size=beam_size,
model_path=model_path,
use_gpu=False,
config_path=config_path,
saveto=saveto,
max_steps=20)
def test_transformer_inference(test_dir, use_gpu=False):
from src.bin import translate
from src.utils.common_utils import GlobalNames
config_path = "./unittests/configs/test_transformer.yaml"
saveto = os.path.join(test_dir, "save")
model_name = test_utils.get_model_name(config_path)
model_path = os.path.join(saveto, model_name + GlobalNames.MY_BEST_MODEL_SUFFIX + ".final")
source_path = "./unittests/data/dev/zh.0"
reference_path = "./unitests/data/dev/en"
batch_size = 3
beam_size = 3
alpha = 0.6
translate.run(model_name=model_name,
source_path=source_path,
reference_path=reference_path,
batch_size=batch_size,
beam_size=beam_size,
model_path=model_path,
use_gpu=False,
config_path=config_path,
saveto=saveto,
max_steps=20,
alpha=alpha)
def test_transformer_train(test_dir):
from src.bin import train_weighted_ocd
config_path = "./unittests/configs/test_odc_loss.yaml"
model_name = test_utils.get_model_name(config_path)
saveto = os.path.join(test_dir, "save")
log_path = os.path.join(test_dir, "log")
valid_path = os.path.join(test_dir, "valid")
train_weighted_ocd.run(model_name=model_name,
config_path=config_path,
saveto=saveto,
log_path=log_path,
valid_path=valid_path,
debug=True)
def test_transformer_train(test_dir, use_gpu=False):
from src.bin import train
config_path = "./unittests/configs/test_transformer.yaml"
model_name = test_utils.get_model_name(config_path)
saveto = os.path.join(test_dir, "save")
log_path = os.path.join(test_dir, "log")
valid_path = os.path.join(test_dir, "valid")
train.run(model_name=model_name,
config_path=config_path,
saveto=saveto,
log_path=log_path,
valid_path=valid_path,
debug=True)
def test_dl4mt_inference(test_dir):
from src.bin import translate
from src.utils.common_utils import GlobalNames
config_path = "./unittests/configs/test_dl4mt.yaml"
saveto = os.path.join(test_dir, "save")
model_name = test_utils.get_model_name(config_path)
model_path = os.path.join(saveto, model_name + GlobalNames.MY_BEST_MODEL_SUFFIX)
source_path = "./unittests/data/dev/zh.0"
batch_size = 3
beam_size = 3
translate.run(model_name=model_name,
source_path=source_path,
batch_size=batch_size,
beam_size=beam_size,
model_path=model_path,
use_gpu=False,
config_path=config_path,
saveto=saveto,
max_steps=20)
def test_transformer_greedy_search(test_dir, use_gpu=False):
from src.bin import translate
from src.utils.common_utils import Constants
config_path = "./unittests/configs/test_transformer.yaml"
saveto = os.path.join(test_dir, "save")
model_name = test_utils.get_model_name(config_path)
model_path = os.path.join(
saveto, model_name + Constants.MY_BEST_MODEL_SUFFIX + ".final")
source_path = "./unittests/data/dev.de"
batch_size = 3
beam_size = 1
alpha = 0.6
translate.run(model_name=model_name,
source_path=source_path,
batch_size=batch_size,
beam_size=beam_size,
model_path=model_path,
use_gpu=False,
config_path=config_path,
saveto=saveto,
max_steps=10,
alpha=alpha)
