search

guanlinchao / bert-dst

BERT-DST: Scalable End-to-End Dialogue State Tracking with Bidirectional Encoder Representations from Transformer
101 stars 45 forks source link

can I use multi gpu? #5

Open guhuawuli opened 4 years ago

guhuawuli commented 4 years ago

The training process need 6 hours in one k80 gpu. Can I use multi GPU? Does that need to change the code ?