def search_word(self, instance, value):
config.print_args(instance,
value,
args_name='instance',
kwargs_name='value')
res = config.dictionary.search_word(value)
print('Search result: {}'.format(res))
# self.rv.data = [{'text':'{} - {}'.format(key, ' // '.join(value))} for key, value in res.items()]
self.rv.data = config.recycleview_data_formatter(res)
return None
type=float,
default=0.0,
help=
"If null_score - best_non_null is greater than the threshold predict null."
)
args = parser.parse_args()
batch_size = args.num_nodes * args.gpu_num_per_node * args.batch_size_per_device
eval_batch_size = args.num_nodes * args.gpu_num_per_node * args.eval_batch_size_per_device
epoch_size = math.ceil(args.train_example_num / batch_size)
num_eval_steps = math.ceil(args.eval_example_num / eval_batch_size)
args.iter_num = epoch_size * args.num_epochs
args.predict_batch_size = eval_batch_size
configs.print_args(args)
def SquadDecoder(data_dir,
batch_size,
data_part_num,
seq_length,
is_train=True):
with flow.scope.placement("cpu", "0:0"):
ofrecord = flow.data.ofrecord_reader(data_dir,
batch_size=batch_size,
data_part_num=data_part_num,
random_shuffle=is_train,
shuffle_after_epoch=is_train)
blob_confs = {}
help="If null_score - best_non_null is greater than the threshold predict null.",
)
args = parser.parse_args()
batch_size = args.num_nodes * args.gpu_num_per_node * args.batch_size_per_device
eval_batch_size = (
args.num_nodes * args.gpu_num_per_node * args.eval_batch_size_per_device
)
device = flow.device(args.device)
epoch_size = math.ceil(args.train_example_num / batch_size)
num_eval_steps = math.ceil(args.eval_example_num / eval_batch_size)
args.iter_num = epoch_size * args.num_epochs
args.predict_batch_size = eval_batch_size
config.print_args(args)
def save_model(module: nn.Module, checkpoint_path: str, name: str):
snapshot_save_path = os.path.join(checkpoint_path, f"snapshot_{name}")
if not os.path.exists(checkpoint_path):
os.makedirs(checkpoint_path)
print(f"Saving model to {snapshot_save_path}")
flow.save(module.state_dict(), snapshot_save_path)
class SquadDecoder(nn.Module):
def __init__(self, data_dir, batch_size, data_part_num, seq_length, is_train=True):
super().__init__()
self.is_train = is_train
from trainer import Trainer
if __name__ == '__main__':
if not args.debug_mode:
import wandb
wandb.init(project=args.project,
name=args.name,
tags=args.tags,
config=args)
train_data = dataset.MDB_Dataset('MusicDelta_80sRock')
test_data = dataset.MDB_Dataset('MusicDelta_80sRock')
else:
train_data = dataset.MDB_Dataset('MusicDelta_80sRock')
test_data = dataset.MDB_Dataset('MusicDelta_80sRock')
print_args(args)
# get_model
if args.model_arc == 'CNN':
model = CNN(hidden_channel_num=10, output_number=4)
else:
raise AssertionError
model = model.to(args.device)
optimizer = torch.optim.Adam(model.parameters(), lr=args.learning_rate)
if not args.debug_mode:
wandb.watch(model)
trainer = Trainer(model, optimizer, args.device, args.debug_mode,
args.test_per_epoch, args.num_epochs, args.weight_path,