horovod · Issues· 407 open
Open on GitHubLocally synced open issues (discussions stay on GitHub)
- #4307
Failed to install horovod in NGC pytorch image
bugUpdated Oct 10, 2025 - #4039
v0.28.1 Version Mismatch with TF 2.12.0. Works with v0.28.0
bugUpdated Sep 23, 2025 - #4290
Doesn't Horovod support TensorFlow >= 2.16.1 ?
bugUpdated Sep 11, 2025 - #4301
Will Horovod support the symmetric memory feature of NCCL 2.27?
enhancementUpdated Aug 12, 2025 - #4073
Failed to install Horovod
bugUpdated Apr 8, 2025 - #4020
Horovod with TensorFlow crashed
bugUpdated Mar 21, 2025 - #4038
Can horovd process more shards than workers
wontfixUpdated Jan 31, 2025 - #4043
NVIDIA CUDA TOOLKIT version to run Horovod in Conda Environment
wontfixUpdated Jan 31, 2025 - #4046
Horovod with Spark - Job Not Distributing Across Worker Nodes
wontfixUpdated Jan 31, 2025 - #4074
Implement Adaptive Batch Size Scaling for Optimized Distributed Training
enhancementUpdated Nov 4, 2024 - #4069
Unable to install horovod on aarch64 platfrom either in host or container
bugUpdated Oct 18, 2024 - #3971
Launching horovod task function was not successful
bugUpdated Oct 17, 2024 - #3483
Horovod cannot be installed together with ML frameworks with pip>=22.0
bugcontribution welcomeUpdated Oct 15, 2024 - #4040
Replace tf.train.SessionRunHook by tf.compat.v1.train.SessionRunHook ?
bugUpdated Oct 8, 2024 - #4067
Respin the Docker container updating the components
enhancementUpdated Aug 16, 2024