Tags give the ability to mark specific points in history as being important
Force allreduce of all gradients in step(), bugfixes
Add compatibility with PyTorch 0.4.1
Support for IBM PowerAI DDL & APIs to restore optimizer state
Critical Bugfix: PyTorch must wait for GPU data before allreduce
Critical Bugfix: non-fused allreduce produces incorrect results
Fix PyTorch master break - use proper THTensor_storage() API
Bugfix mpi4py: Create a private MPI communicator
Collective control plane & other low latency improvements
Bugfix: `python setup.py install` requires cffi
Support TensorFlow Optimizers that override compute_gradients/apply_gradients
Avoid deadlock if worker crashes with an exception
New major release
Add Keras ImageNet training example and LearningRateScheduleCallback