



Installing pip dependencies: - Ran pip subprocess with arguments: Writing config: C:\Users\asm90\Anaconda3\envs\DEEPLABCUT\etc\jupyterĮnabling notebook extension jupyter-js-widgets/extension….\ Config option kernel_spec_manager_class not recognized by EnableNBExtensionApp.Įnabling notebook extension nb_conda/main…Ĭonfig option kernel_spec_manager_class not recognized by EnableServerExtensionApp. (base) C:\Users\asm90\DeepLabCut\conda-environments>conda env create -f DEEPLABCUT.yamlĬollecting package metadata (repodata.json): doneĮxecuting transaction: - Enabling nb_conda_kernels…ĬONDA_PREFIX: C:\Users\asm90\Anaconda3\envs\DEEPLABCUT (base) C:\Windows\system32>cd C:\Users\asm90\DeepLabCut\conda-environments You can list all discoverable environments with conda info -envs. (base) C:\Windows\system32>activate DEEPLABCUTĬould not find conda environment: DEEPLABCUT I would be grateful if anyone can take a look at the following and identify what’s causing the DLC anaconda environment set-up file (DEEPLABCUT.yaml) to set me up with tensorflow 2.2.0 as opposed to a version that is compatible with cuda 11 (tensorflow >2.4.0). I have tried to force the DLC conda environment to accept tensorflow 2.6.0 but run into numpy version issues. Still it solves the environment with tensorflow 2.2.0 and use of tensorflow by deeplabcut returns errors stating that cuda 10 related files are missing. And removed and then re-created the DLC anacoda environment. I then uninstalled cuda 10 - so that only cuda 11 is present on the device. Which is not compatible with cuda 11 but rather attempts to load cuda 10.
#Anaconda cuda toolkit driver#
I then git clone (or manually download) the anaconda environment file DEEPLABCUT.yaml and run conda env create -f DEEPLABCUT.yamlĭespite starting with the latest version of CUDA - and a compatible NVIDIA driver - when the DLC anaconda environment is created I end up with tensorflow 2.2.0. Test (in base anaconda environment) whether tensorflow finds an available GPU (tf.test.is_gpu_available): returns True
