Description
🐛 Bug
Describe the bug
When trying to install the nightly builds of PyTorch and TorchText with CUDA, the PyTorch install is coerced into a CPU-only one.
Running conda install pytorch torchtext cudatoolkit=11.3 -c pytorch-nightly will try to install pytorch==pytorch-nightly/linux-64::pytorch-1.11.0.dev20220126-py3.9_cpu_0.
To Reproduce
Create a new conda environment, then try to install PyTorch and TorchText nightly builds with CUDA support.
conda create --name torchtext
conda activate torchtext
conda install pytorch torchtext cudatoolkit=11.3 -c pytorch-nightly
Expected behavior
Expected behavior should match the equivalent install process for e.g. TorchVision.
conda create --name torchvision
conda activate torchvision
conda install pytorch torchvision cudatoolkit=11.3 -c pytorch-nightly
This will try to install pytorch==pytorch-nightly/linux-64::pytorch-1.11.0.dev20220126-py3.9_cuda11.3_cudnn8.2.0_0, as expected (see also the below screenshots).
Screenshots
List of packages to be installed by conda install pytorch torchtext -c pytorch-nightly:
List of packages to be installed by conda install pytorch torchtext -c pytorch-nightly:
Environment
Results of environment collection script:
python collect_env.py
Collecting environment information...
PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
ROCM used to build PyTorch: N/A
OS: Ubuntu 20.04.2 LTS (x86_64)
GCC version: (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0
Clang version: Could not collect
CMake version: version 3.16.3
Libc version: glibc-2.29
Python version: 2.7.18 (default, Mar 8 2021, 13:02:45) [GCC 9.3.0] (64-bit runtime)
Python platform: Linux-5.4.0-81-generic-x86_64-with-Ubuntu-20.04-focal
Is CUDA available: N/A
CUDA runtime version: Could not collect
GPU models and configuration:
GPU 0: Quadro GP100
GPU 1: Quadro GP100
Nvidia driver version: 470.57.02
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Versions of relevant libraries:
[pip] Could not collect
[conda] # packages in environment at /private/home/ebs/miniconda/envs/torchtext: