Top tensor2tensor issues

Why does transformer use noam as the learning rate decay scheme?

anglil picture anglil·6 Sep 2017·17 Comments

[Tuning] Results are GPU-number and batch-size dependent

question
vince62s picture vince62s·27 Nov 2017·81 Comments
👍10

AttributeError: module 'tensorflow' has no attribute 'flags'

birdmw picture birdmw·26 Nov 2019·6 Comments

Why add positional embedding instead of concatenate?

Akella17 picture Akella17·30 May 2019·5 Comments
👍27

how to reproduce "one model to learn them all" results?

zzkszzks picture zzkszzks·21 Jun 2017·5 Comments
👍4

Export/Server tensor2tensor as a .pb file instead of .pbtxt

question
NehaPathapati picture NehaPathapati·6 Jun 2018·8 Comments

*help* how to just use the transformer encoder layer?

doubler picture doubler·17 May 2018·13 Comments
👍2

Inference speed for Transformer Vs GNMT. Transformer is slower

question
efeiefei picture efeiefei·27 Feb 2018·23 Comments
👍1

ERROR:tensorflow:Model diverged with loss = NaN during traning translation model

yudianer picture yudianer·11 Feb 2018·4 Comments

Unable to reproduce WMT En2De results

question
edunov picture edunov·21 Sep 2017·52 Comments
Next issues