search

playma / LCSTS2.0-clean

7 stars 1 forks source link

How to set transformer's hyperparameter? #2

Open wanwanaa opened 5 years ago

wanwanaa commented 5 years ago

hi playma, i am trying to use HWC+transformer, but the results is unsatisfactory. Could you tell me what hyperparameres you use(batch size, sequence length)? How many gpu are use when training transformer? Thanks!

playma commented 5 years ago

I use transformer model implemented by OpenNMT, and there are some script with hyperparameter in below repo. https://github.com/playma/OpenNMT-py/tree/master/LCSTS2.0-clean_script