def main():
args = get_train_args()
model = init_train_env(args, tbert_type='siamese2')
valid_examples = load_examples(args.data_dir, data_type="valid", model=model, num_limit=args.valid_num,
overwrite=args.overwrite)
train_examples = load_examples(args.data_dir, data_type="train", model=model, num_limit=args.train_num,
overwrite=args.overwrite)
train(args, train_examples, valid_examples, model, train_iter_method=train_with_neg_sampling)
logger.info("Training finished")
def main():
args = get_train_args()
model = init_train_env(args, tbert_type='siamese2')
train_dir = os.path.join(args.data_dir, "train")
valid_dir = os.path.join(args.data_dir, "valid")
train_examples = load_examples(train_dir, model=model, num_limit=args.train_num)
valid_examples = load_examples(valid_dir, model=model, num_limit=args.valid_num)
train(args, train_examples, valid_examples, model, train_with_neg_sampling)
logger.info("Training finished")
def main():
args = get_train_args()
model = init_train_env(args, tbert_type='single')
valid_examples = load_examples(args.data_dir,
data_type="valid",
model=model,
num_limit=args.valid_num,
overwrite=args.overwrite)
train_examples = load_examples(args.data_dir,
data_type="train",
model=model,
num_limit=args.train_num,
overwrite=args.overwrite)
train(args, train_examples, valid_examples, model, train_single_iteration)
logger.info("Training finished")
def main():
args = get_rnn_train_args()
if args.is_no_padding:
args.gradient_accumulation_steps = 1
args.per_gpu_eval_batch_size = 1
args.per_gpu_train_batch_size = 1
args.logging_steps = args.logging_steps * 10
args.exp_name = "{}_{}".format(
args.exp_name,
datetime.datetime.now().strftime("%m-%d-%H-%M-%S"))
# embd_info = create_emb_layer("./we/glove.6B.300d.txt")
embd_info = load_embd_from_file(args.embd_file_path)
if args.local_rank == -1 or args.no_cuda:
device = torch.device("cuda" if torch.cuda.is_available()
and not args.no_cuda else "cpu")
args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
else: # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
torch.cuda.set_device(args.local_rank)
device = torch.device("cuda", args.local_rank)
torch.distributed.init_process_group(backend="nccl")
args.n_gpu = 1
args.device = device
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
level=logging.INFO if args.local_rank in [-1, 0] else logging.WARN,
)
if args.local_rank not in [-1, 0]:
# Make sure only the first process in distributed training will download model & vocab
torch.distributed.barrier()
model = RNNTracer(hidden_dim=args.hidden_dim,
embd_info=embd_info,
embd_trainable=args.is_embd_trainable,
max_seq_len=args.max_seq_len,
is_no_padding=args.is_no_padding,
rnn_type=args.rnn_type)
if args.local_rank == 0:
# Make sure only the first process in distributed training will download model & vocab
torch.distributed.barrier()
model.to(args.device)
model.device = args.device
if args.fp16:
try:
import apex
apex.amp.register_half_function(torch, "einsum")
except ImportError:
raise ImportError(
"Please install apex from https://www.github.com/nvidia/apex to use fp16 training."
)
train_dir = os.path.join(args.data_dir, "train")
valid_dir = os.path.join(args.data_dir, "valid")
train_examples = load_examples_for_rnn(train_dir,
model=model,
num_limit=args.train_num)
valid_examples = load_examples_for_rnn(valid_dir,
model=model,
num_limit=args.valid_num)
logger.info("Training started")
train(args, train_examples, valid_examples, model, train_rnn_iter)
logger.info("Training finished")