train.py: renaming function, removing dataparallel

experiments
rafaelvalle 2018-11-27 18:04:12 -08:00
parent 3045ba125b
commit f06063f746
1 changed files with 1 additions and 3 deletions

View File

@ -84,9 +84,7 @@ def load_model(hparams):
model.decoder.attention_layer.score_mask_value = float(finfo('float16').min) model.decoder.attention_layer.score_mask_value = float(finfo('float16').min)
if hparams.distributed_run: if hparams.distributed_run:
model = DistributedDataParallel(model) model = apply_gradient_allreduce(model)
elif torch.cuda.device_count() > 1:
model = DataParallel(model)
return model return model