def get_model(config, args, train_dataset, device):
# Pass vocabulary to construct Embedding layer.
encoder = Encoder(config["model"], train_dataset.vocabulary)
decoder = Decoder(config["model"], train_dataset.vocabulary)
print("Encoder: {}".format(config["model"]["encoder"]))
print("Decoder: {}".format(config["model"]["decoder"]))
# New: Initializing word_embed using GloVe
if "glove_npy" in config["dataset"]:
encoder.word_embed.weight.data = torch.from_numpy(
np.load(config["dataset"]["glove_npy"]))
print("Loaded glove vectors from {}".format(
config["dataset"]["glove_npy"]))
# Share word embedding between encoder and decoder.
if encoder.word_embed and decoder.word_embed:
decoder.word_embed = encoder.word_embed
# Wrap encoder and decoder in a model.
model = EncoderDecoderModel(encoder, decoder).to(device)
if -1 not in args.gpu_ids:
model = nn.DataParallel(model, args.gpu_ids)
return model
val_dataset,
batch_size=config["solver"]["batch_size"],
num_workers=args.cpu_workers,
shuffle=True,
)
# Pass vocabulary to construct Embedding layer.
encoder = Encoder(config["model"], val_dataset.vocabulary)
decoder = Decoder(config["model"], val_dataset.vocabulary)
print("Encoder: {}".format(config["model"]["encoder"]))
print("Decoder: {}".format(config["model"]["decoder"]))
# Share word embedding between encoder and decoder.
if args.load_pthpath == "":
print('load glove')
decoder.word_embed = encoder.word_embed
glove = np.load('data/glove.npy')
encoder.word_embed.weight.data = torch.tensor(glove)
# Wrap encoder and decoder in a model.
model = EncoderDecoderModel(encoder, decoder).to(device)
if -1 not in args.gpu_ids:
model = nn.DataParallel(model, args.gpu_ids)
# =============================================================================
# SETUP BEFORE TRAINING LOOP
# =============================================================================
start_time = datetime.datetime.strftime(datetime.datetime.utcnow(), '%d-%b-%Y-%H:%M:%S')
sparse_metrics = SparseGTMetrics()
ndcg = NDCG()