Tags: kashif/horovod
Toggle v0.16.4's commit message
Improved optimizer.skip_synchronize() API for NVIDIA AMP
Toggle v0.16.3's commit message
Intel MLSL support, improvements to gradient clipping & NVIDIA AMP pe…
…rformance, bugfixes
Toggle v0.16.2's commit message
Apache MXNet 1.4.1 compatibility, improvements to coordination perfor…
…mance at ultra-large scale, stall message improvements, bugfixes
Toggle v0.16.1's commit message
Toggle v0.16.0's commit message
PySpark, Apache MXNet, autotuning, TensorFlow eager execution, mixed-…
…precision & embedding improvements
Toggle v0.15.2's commit message
Force allreduce of all gradients in step(), bugfixes
Toggle v0.15.1's commit message
FP16 hierarchical allreduce
Toggle v0.15.0's commit message
PyTorch 1.0, TF-Keras, FP16 ops on GPU
Toggle v0.14.1's commit message
Parallelized hierarchical allreduce
Toggle v0.14.0's commit message
Support for the upcoming PyTorch release
You can’t perform that action at this time.