fix mask bug in star-transformer

fix the bug described in #138 . Thank @wlhgtc for bug reporting and pr.

fastNLP/modules/encoder/star_transformer.py

@@ -46,6 +46,7 @@ class StarTransformer(nn.Module):
             return f(x.permute(0, 2, 3, 1)).permute(0, 3, 1, 2)
 
         B, L, H = data.size()
+        mask = (mask == 0) # flip the mask for masked_fill_
         smask = torch.cat([torch.zeros(B, 1, ).byte().to(mask), mask], 1)
 
         embs = data.permute(0, 2, 1)[:,:,:,None] # B H L 1