 Max H. Gerlach
|
1d217b5994
Bump version to v0.28.1 (#3942)
|
1 年之前 |
|
Max H. Gerlach
|
19093eaebe
Bump version to 0.28.0 (#3910)
|
1 年之前 |
|
romerojosh
|
004fd0d93f
Update with_device functions in MXNet and PyTorch to skip unnecessary cudaSetDevice calls (#3912)
|
1 年之前 |
|
Enrico Minack
|
3a9bf1ba1f
Upgrade CI test frameworks and fix linking NCCL 2.12+ (#3846)
|
1 年之前 |
|
Nicolas Castet
|
2ef8ff92f5
Re-enabling new keras optimizers (#3860)
|
1 年之前 |
|
Pei-Lun Liao
|
2af9000a24
Force tf.logical_and in hvd allreduce condition running on CPU (#3885)
|
1 年之前 |
|
Ata Fatahi
|
f356349204
TF: Add get_local_and_global_gradients() to PartialDistributedGradentTape (#3859)
|
1 年之前 |
|
Max H. Gerlach
|
687eb29d25
Build: Link CUDA runtime library statically by default (#3867)
|
1 年之前 |
|
Max H. Gerlach
|
2ecd714b08
Fix ROCm build for recent Reducescatter changes (#3839)
|
1 年之前 |
|
Max H. Gerlach
|
b46c5b7085
Reducescatter: Allocate output tensors before enqueuing the operation (#3824)
|
1 年之前 |
|
Max H. Gerlach
|
bfaca90d5c
Bump version to 0.27.0 (#3832)
|
1 年之前 |
|
Lee Yang
|
88ecd061a3
support custom data loaders in TorchEstimator (#3787)
|
1 年之前 |
|
Max H. Gerlach
|
25e0f6476a
Add prescale_factor and postscale_factor for Reducescatter (#3815)
|
1 年之前 |
|
Li Jiang
|
ede98b82ab
Support pinning to local rank GPU index in Spark estimators (#3737)
|
1 年之前 |
|
Nicolas Castet
|
305280def3
Fix broken TF DistributedOptimizer with Keras 2.11+ (#3822)
|
1 年之前 |
|
romerojosh
|
85619c6851
Several fixes for allreduce and grouped allreduce handling of tf.IndexedSlices. (#3813)
|
1 年之前 |
|
romerojosh
|
f99dfbf84d
Handle tf.IndexedSlices types when scaling local gradients in TF. (#3786)
|
1 年之前 |
|
Max H. Gerlach
|
b9d0e14801
Pad fused GPU Allgathers for better memory alignment (#3727)
|
2 年之前 |
|
Enrico Minack
|
80e4894860
Update CHANGELOG.md from release (#3740)
|
2 年之前 |
|
Enrico Minack
|
34604870ea
Bum version to 0.26.1 (#3745)
|
2 年之前 |
|
Travis Addair
|
b601ec9a2c
Fixed packaging import to occur after install_requires (#3741)
|
2 年之前 |
|
Enrico Minack
|
c638dcec97
Bump version to 0.26.0 (#3731)
|
2 年之前 |
|
chongxiaoc
|
4e6eaa3780
Keras: Support only legacy optimizers in Keras 2.11+ (#3725)
|
2 年之前 |
|
Ata Fatahi
|
4e82ad0144
Support registering local variables for BroadcastGlobalVariablesCallback (#3703)
|
2 年之前 |
|
Max H. Gerlach
|
37a6d83dd9
Fix reducescatter() and grouped_reducescatter() to raise clean exceptions for scalar inputs (#3699)
|
2 年之前 |
|
Ata Fatahi
|
4f723bb6a5
TF: Add register_local_var to distributed optimizers and gradient aggregators (#3695)
|
2 年之前 |
|
romerojosh
|
25ed80371c
Enable use of native ncclAvg op for NCCL allreduces. (#3646)
|
2 年之前 |
|
Max H. Gerlach
|
daf0f4111a
Build: Fix finding nvcc (if not in $PATH) with older versions of CMake (#3682)
|
2 年之前 |
|
Max H. Gerlach
|
81c4f9ab5f
Revert "TF: Add register_local_var to distributed optimizers and gradient aggregators (#3663)" (#3686)
|
2 年之前 |
|
Ata Fatahi
|
dfe5b8ee27
TF: Add register_local_var to distributed optimizers and gradient aggregators (#3663)
|
2 年之前 |
