 Lee Yang
|
a658d0f531
Add custom data loading (e.g. NVTabular) in KerasEstimator (#3603)
|
2 年之前 |
|
chongxiaoc
|
001260aeb3
Spark: adapt Petastorm 0.12.0 changes to speedup data shuffling (#3665)
|
2 年之前 |
|
Serena Ruan
|
da4bef49ec
Spark/Lightning: add params of edit fields for petastorm datamodule (#3651)
|
2 年之前 |
|
Vignesh Kothapalli
|
87b5fc7c11
Support uint8 and int8 allreduce in tensorflow (#3649)
|
2 年之前 |
|
Ata Fatahi
|
35182d764f
TF: Add PartialDistributedGradientTape API (#3643)
|
2 年之前 |
|
romerojosh
|
a9a8d5770f
Fix race condition in PyTorch allocation handling. (#3639)
|
2 年之前 |
|
Vignesh Kothapalli
|
8c30b6e690
fix the optimizer iteration increment logic when gradient accumulation is enabled (#3631)
|
2 年之前 |
|
romerojosh
|
e1bf78aafc
Add register_local_source and use_generic_names funtionality to DistributedGradientTape for TF. (#3628)
|
2 年之前 |
|
Max H. Gerlach
|
8f450abc19
Add hvd.grouped_allgather and hvd.grouped_reducescatter (#3594)
|
2 年之前 |
|
Max H. Gerlach
|
757883be5c
Add support for batched memory copies in GPUReducescatter (#3621)
|
2 年之前 |
|
Max H. Gerlach
|
da336f59aa
Update Eigen submodule to fix build on macOS with aarch64 (#3619)
|
2 年之前 |
|
Enrico Minack
|
48e0affcba
Bump version to 0.25.0 (#3581)
|
2 年之前 |
|
Max H. Gerlach
|
176251b244
Add NVTX op tracing for Reducescatter, make some base class destructors virtual (#3574)
|
2 年之前 |
|
Enrico Minack
|
32b5d3ecbb
Add Horovod job to run Tensorflow Data Service (#3525)
|
2 年之前 |
|
Max H. Gerlach
|
1b3452f686
Update docs and changelog for PR #3558 (#3571)
|
2 年之前 |
|
Amog Kamsetty
|
a0cd0af215
Make `HorovodVersionMismatchError` subclass `ImportError` (#3549)
|
2 年之前 |
|
chongxiaoc
|
7707267a4b
Spark/Lightning: add missing tranform_spec for Petastorm datamodule (#3543)
|
2 年之前 |
|
Enrico Minack
|
6285f3f8c4
Rework network interfaces args in horovod.run and horovodrun (#3506)
|
2 年之前 |
|
chongxiaoc
|
0e0c7dcd3f
[Spark]: expose random seed as an optional parameter (#3517)
|
2 年之前 |
|
Enrico Minack
|
9af980ef46
Bump version to 0.24.3 (#3518)
|
2 年之前 |
|
romerojosh
|
f9d7f77d01
Make TensorFlow output allocations asynchronous when using NCCL backend. (#3464)
|
2 年之前 |
|
Nicolas Castet
|
b4db4052d3
Fallback to NCCL shared lib if static one is not found (#3500)
|
2 年之前 |
|
Travis Addair
|
4cdb671399
Bump version to 0.24.2 (#3467)
|
2 年之前 |
|
Nicolas Castet
|
4111d6b7ae
Fix ignored cuda arch flags (#3462)
|
2 年之前 |
|
Max H. Gerlach
|
8a12dc9797
Add Reducescatter op (NCCL, MPI, Gloo) (#3299)
|
2 年之前 |
|
chongxiaoc
|
e02bdca74d
[Setup] Require fsspec >= 2010.07.0 (#3451)
|
2 年之前 |
|
Enrico Minack
|
ebd1350985
Bump version to 0.24.1 (#3446)
|
2 年之前 |
|
Max H. Gerlach
|
8147526523
Help CMake find CUDA even if nvcc is not in $PATH (#3444)
|
2 年之前 |
|
Enrico Minack
|
b089df66a2
Bump version to 0.24.0 (#3433)
|
2 年之前 |
|
chongxiaoc
|
642a6b3018
[TF - Fix] Fix imports from tensorflow.python.keras with tf.__version__ >= 2.6.0 (#3403)
|
2 年之前 |
