use_gpu = torch.cuda.is_available()
if use_gpu:
torch.cuda.set_device(gpu_to_use)
if not os.path.isdir(output_dir):
os.mkdir(output_dir)
save_params = (os.path.join(output_dir,
model_name), os.path.join(output_dir, log_name))
rnn = EncoderDecoder(hidden_dim,
otu_handler,
num_lstms,
use_gpu,
LSTM_in_size=num_strains,
use_attention=use_attention)
rnn.do_training(inp_slice_len,
target_slice_len,
batch_size,
num_epochs,
learning_rate,
samples_per_epoch,
teacher_force_frac,
weight_decay,
save_params=save_params,
use_early_stopping=use_early_stopping,
early_stopping_patience=early_stopping_patience,
inp_slice_incr_frequency=inp_slice_incr_frequency,
target_slice_incr_frequency=target_slice_incr_frequency)
"--vocab",
type=str,
help="The file with the vocab characters.")
args = parser.parse_args()
input_dir = args.data
vocab_file = args.vocab
files = [
os.path.join(input_dir, f) for f in os.listdir(input_dir)
if f.endswith('.csv')
]
TH = TweetHandler(files, vocab_file)
TH.set_train_split()
TH.remove_urls()
if not os.path.isdir(output_dir):
os.mkdir(output_dir)
save_params = (os.path.join(output_dir,
model_name), os.path.join(output_dir, log_name))
enc = EncoderDecoder(hidden_dim, TH, num_lstms)
enc.do_training(seq_len,
batch_size,
num_epochs,
learning_rate,
samples_per_epoch,
teacher_force_frac,
slice_incr_frequency=slice_incr_frequency,
save_params=save_params)