def get_random_object() -> str:
return randomname.get_name(sep=' ')
def default_unique_experiment_name(cls, v, values): # noqa: D102, N805
return v or values["experiment_name"] + "_" + randomname.get_name()
def test_get_name():
name = randomname.get_name('music_theory', 'cats').split('-', 1)
assert len(name) > 1
assert name[0] in randomname.util.get_groups_list('a/music_theory')
assert name[1] in randomname.util.get_groups_list('n/cats')
assert 'asdf' in randomname.util.get_groups_list(['n/cats', 'asdf'])
options = [line.rstrip() for line in f]
else:
options = [env_option]
return [_get_env_variable(option) for option in options]
def _parse_env_options(env_options):
return dict(
[j for i in [_get_env_variables(k) for k in env_options] for j in i])
@cli.command(help="Stages and starts a Mendix application.")
@click.option(
"-n",
"--name",
default=randomname.get_name(),
help="Sets the name of the application.",
)
@click.option(
"-p",
"--password",
help="Sets the adminstrator password for the application.",
)
@click.option(
"-e",
"--env",
multiple=True,
help=
"Sets an environment variable (KEY=VALUE) for the application. Providing a file with environment variables and multiple options are allowed.",
)
@click.option(
from config import config
from loader import GenreDataLoader
from networks import GenreAutoencoder, GenreDecoder
bert_config_name = "prajjwal1/bert-mini"
BATCH_SIZE = 64
bert_config = BertConfig.from_pretrained(bert_config_name)
PATH = f"genre_autoencoder-{bert_config.hidden_size}"
lp_path = os.path.join(config.get("learning_progress_path"), PATH)
TRAIN_STEPS = int(1e5)
if __name__ == "__main__":
run_name = randomname.get_name()
model_dump_path = os.path.join(lp_path, "model", run_name)
Path(model_dump_path).mkdir(parents=True, exist_ok=True)
data_loader = GenreDataLoader(
batch_size=BATCH_SIZE,
meta_data_path="data/album_data_frame.json",
tokenizer_path=bert_config_name,
)
num_labels = data_loader.get_number_of_classes()
vocab_size = len(data_loader.tokenizer.get_vocab())
encoder = BertForMaskedLM.from_pretrained(bert_config_name)
encoder.bert = BertModel.from_pretrained(bert_config_name,
add_pooling_layer=True)
decoder = GenreDecoder(input_dim=bert_config.hidden_size,