Yuefeng Zhou comments

Repositories
Issues
Comments

Results 4 comments of


                                            Yuefeng Zhou

how can the parameter server stop itself?

It is safe to kill the ps process after your training is done (and your checkpoint is saved as well). Do you have a specific concern?

RFC: Sparse Domain Isolation for Supporting large-scale Sparse Weights Training.

@byronyi If we are going to contribute to addon first, do we need a RFC here?

RFC: Sparse Domain Isolation for Supporting large-scale Sparse Weights Training.

I think TensorFlow can provide a way to extend optimizers so that you can extend existing optimizers to handle your sparse weights.

Enable multiple input pipelines with a single machine, MultiWorkerMirroredStrategy's experimental_distribute_datasets_from_function

Hello @jeisinge , the API doc says `Each replica on that worker will dequeue one batch of inputs from the local Dataset (i.e. if a worker has two replicas, two...