colmantse

Results 4 comments of colmantse

Hanging when running transformer in distributed setting

I met the same problem as well. interestingly tho, estimator are supposed to handle the start_std_server in tf.estimator.training but for some reasons, it is not able to in current t2t.

No module named tensor2tensor

hey can you try import something like: from tensor2tensor.layers import common_layers

No module named tensor2tensor

hi, I suppose it might be path error, you might want to locate the tensor2tensor folder and add it to your path via export. maybe the t2t guys know better.

Change kwargs=None optional parameter to **kwargs

sorry has it been merged yet? facing the same issue.