why not set bn_training=False when test net on x_qry in finetunning phase

[boy-be-ambitious]

in finetunning phase why you set bn_training=True when test net on x_qry ?

line 178 、188 and 204 in meta,py

[jaekyeom]

I'm not the author, but https://github.com/cbfinn/maml/issues/9 might answer your question.

[liubai01]

I have faced with the same problem. It sounds better if we screen off the running batch mean/var estimation when testing(a.k.a. fine-turning). I will try it.