Environment:

Framework: (TensorFlow, Keras)
Framework version:
tensorflow 1.14.0
tensorflow-estimator 1.14.0
tensorflow-serving-api 1.14.0
Keras 2.2.4
Keras-Applications 1.0.8
Keras-Preprocessing 1.1.0
Horovod version:
horovod 0.18.1
MPI version:
(tensorflow_p36) ubuntu@ip-172-31-38-183:~$ mpirun --version
mpirun (Open MPI) 4.0.1
CUDA version:
CUDA Version 10.0.130

NCCL version

Train a simple NER tagger for Swedish trained for instance over this dataset.

For this task, we need to adapt the NLPTaskDataFetcher for the appropriate Swedish dataset and train a simple model using Swedish word embeddings. How to train a model is [illustrated here](https://github.com/zalandoresearch/flair/blob/master/resources/docs/TUTORIAL_TRAI

If you use the min_count parameter of the Vocabulary, but you specify a namespace that does not exist, the vocabulary creation will just silently proceed. It'd be great if it could error in this case, perhaps by popping off the namespaces in min_count and erroring if any are left at the end of vocab creation (would probably go at the end of https://github.com/allenai/allennlp/blob/master/all

Support for storing large tensor values in external files was introduced in #678, but AFAICT is undocumented.

This is a pretty important feature, functionally, but it's also important for end users who may not realise that they need to move around more than just the *.onnx file.

I would suggest it should be documented in IR.md, and perhaps there are other locations from which it could be s

Originated as a forum post

Let's add a tutorial demonstrating how to serve trained models in C++ using torch.jit.save() and torch::jit::load(). I believe there is some subtlety requiring a torch.jit.trace() to wrap poutine.trace and poutine.replay logic for SVI mod

We can currently send and get tensors, we would like to have the ability to directly fetch datasets with all the data inputs and targets in an easy way.

Things to do:

• Change sy.BaseDataset to inherit from AbstractObject to have tagging and description and to be registrable by workers: this allows datasets to be referenced for the search functionality
• Extend BaseDataset wi