v1.8.0
alexlang74
released this
14 Dec 08:52
·
80 commits
to main
since this release
Open-CE Version 1.8.0
This is release 1.8.0 of Open Cognitive Environment (Open-CE). See #706 for detailed release contents.
Package Versions
A release of Open-CE consists of the environment files within the open-ce
repository and a collection of feedstock repositories. The feedstock repositories contain recipes for various python packages. The following packages (among others) are part of this release:
Package | Version |
---|---|
dali | 1.19.0 |
deepspeed | 0.7.5 |
liblightgbm | 3.3.3 |
av | 10.0.0 |
bazel | 5.1.1 |
boost_mp11 | 1.76.0 |
cmdstan | 2.29.2 |
cmdstanpy | 1.0.7 |
cudatoolkit | 11.2.2 and 11.4.4 |
cudatoolkit-dev | 11.2.2 and 11.4.4 |
cudnn | 8.3.0.98 |
datasets | 2.7.1 |
dm-tree | 0.1.5 |
grpc-cpp | 1.41.0 |
gtest | 1.10.0 |
horovod | 0.26.1 |
jpeg-turbo | 2.1.4 |
keras | 2.10.0 |
libdate | 3.0.1 |
libflac | 1.3.3 |
libiconv | 1.16 |
libsndfile | 1.0.31 |
libsolv | 0.7.19 |
magma | 2.6.1 |
mamba | 1.0.0 |
libmamba | 1.0.0 |
nccl | 2.12.7 |
numactl | 2.0.12 |
onnx | 1.12 |
onnxconverter-common | 1.13.0 |
onnxmltools | 1.11.1 |
onnxruntime | 1.13.1 |
libopencv | 4.6.0 |
openblas | 0.3.21 |
openmpi | 4.1.4 |
optional-lite | 3.4.0 |
orbit-ml | 1.1.1 |
orc | 1.8.0 |
arrow-cpp | 10.0.0 |
prophet | 1.1.1 |
pybind11-abi | 4 |
pytorch-base | 1.13.0 for CUDA 11.4 and cpu |
pytorch-base | 1.10.2 for CUDA 11.2 |
pytorch-lightning | 1.8.2 |
pyDeprecate | 0.3.2 |
pytorch_geometric | 2.1.0 |
pytorch-lightning-bolts | 0.6.0 |
pytorch_scatter | 2.0.8 |
pytorch_sparse | 0.6.10 |
torchmetrics | 0.8.1 |
ray_all | 2.0.1 |
ray-tune | 2.0.1 |
reproc | 14.2.3 |
safeint | 3.0.26 |
sentencepiece | 0.1.96 |
skl2onnx | 1.13 |
spdlog | 1.9.2 |
tensorboard-data-server | 0.6.1 |
tensorboard | 2.10.1 |
tensorflow-addons | 0.18.0 |
tensorflow-datasets | 4.7.0 |
tensorflow-estimator | 2.10.0 |
tensorflow-base | 2.10.1 |
tensorflow-hub | 0.12.0 |
tensorflow-io | 0.27.0 |
tensorflow-io-gcs-filesystem | 0.27.0 |
tensorflow-metadata | 1.11.0 |
tensorflow-model-optimization | 0.7.3 |
tensorflow-probability | 0.18.0 |
tensorflow-text | 2.10.0 |
tensorflow-serving | 2.10.1 |
termcolor-cpp | 2.0.0 |
tf2onnx | 1.13.0 |
torchtext-base | 0.14.0 for CUDA 11.4 and cpu |
torchtext-base | 0.11.2 for CUDA 11.2 |
torchvision-base | 0.14.0 for CUDA 11.4 and cpu |
torchvision-base | 0.11.3 for CUDA 11.2 |
typeguard | 2.12.0 |
libxgboost | 1.7.1 |
xgboost | 1.7.1 |
yaml-cpp | 0.6.3 |
apache-beam | 2.42 |
This release of Open-CE supports NVIDIA's CUDA versions 11.2,11.4 as well as Python 3.8,3.9,3.10.
Getting Started
To get started with this release, see the main readme
Notes
- CUDA support
- This is the last major OpenCE release to support CUDA 11.2.
- CUDA 11.2 limitations:
- The OpenCE recipes build CUDA 11.2 packages only for Python 3.8 and 3.9, not for 3.10
- The PyTorch OpenCE recipe builds PyTorch 1.10. PyTorch 1.13 is only built for CPU and CUDA 11.4.
Known Issues
The following pip check failures are known:
deepspeed 0.7.5+06e00f61 requires ninja, which is not installed.
skl2onnx 1.13 has requirement scikit-learn<=1.1.1, but you have scikit-learn 1.1.3.
pyro-ppl 1.8.3 has requirement torch>=1.11.0, but you have torch 1.10.2.
arviz 0.11.2 has requirement typing-extensions<4,>=3.7.4.3, but you have typing-extensions 4.4.0.
multiprocess 0.70.12.2 has requirement dill>=0.3.4, but you have dill 0.3.1.1.