Jacob Eisinger
Jacob Eisinger
It is interesting to note that we were able to use mixed-precision training before with TF 1.15 and TF Estimator. With TF 2.3 and Keras, we are seeing this same...
We also ran into this issue as well. We solved it by splitting the corpus into ~ 100MB files and then running them as separate processes through a caller script...
We are taking about 25 minutes for one file --- obviously, this is a embarassingly paralleizable task. Also, we adjust the parameters a bit due to having a very large...
We also are running into this peculiarity around the comma and are wondering if this is unintentional or intentional. We noticed that SentencePiece has a function of `EncodeAsIds()` which encodes...
FYI - the reason for the tokenization is listed at https://github.com/google-research/bert/blob/master/README.md#tokenization . Please disregard my previous comment.
Does PR #108 support Spark's `.option()` API?
Any update here? Also, is it possible to enable JIT/XLA as well like https://github.com/tensorflow/serving/issues/1515 ?
Any update here?
Different issue. https://github.com/keras-team/tf-keras/issues/418 is for the lookup. This is for the embedding. Further, embeddings work with XLA - just not with XLA *and* multi gpu training. https://github.com/tensorflow/tensorflow/issues/43687 is similar, but,...
Howdy @nikitamaia, great questions. I should have provided additional context. First, I am working through Python/TF 2 as quickly as I possibly can --- if I say something incorrectly, please...