UnsatisfiableError: The following specifications were found Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.Ĭollecting package metadata (repodata.json): doneįound conflicts! Looking for incompatible packages. Solving environment: failed with initial frozen solve. But by implementing conda install pytorch=1.8.1 torchvision=0.9.0 torchaudio=0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge, I got the error like belowĬollecting package metadata (current_repodata.json): done Src = ScriptRunConfig(source_directory=script_folder,Īctually I found that in order to use A100, pytoch version should be 1.8.1+cu111. Provisioning_config = AmlCompute.provisioning_configuration(vm_size=vm_size,Įnv = Environment.load_from_directory(path="./.azureml6/")Įxp = Experiment(workspace=ws,name=experiment_name)Ĭommand = "pwd & pip install azure-storage-blob & python main.py" If compute_target and type(compute_target) is AmlCompute: Vm_size = os.environ.get("AML_COMPUTE_CLUSTER_SKU", gpu_name)Ĭompute_target = ws.compute_targets Print(ws.name, ws.location, ws.resource_group, sep='\t')Ĭompute_name = os.environ.get("AML_COMPUTE_CLUSTER_NAME", cluster_name)Ĭompute_min_nodes = os.environ.get("AML_COMPUTE_CLUSTER_MIN_NODES", 0)Ĭompute_max_nodes = os.environ.get("AML_COMPUTE_CLUSTER_MAX_NODES", 4) Then in order to create job to computing cliuster I implemented below #A100verĮxperiment_name = 'speaker_identification_training_A100' I implemented conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch, and export to yml file. Warnings.warn(incompatible_device_warn.format(device_name, capability, " ".join(arch_list), device_name)) Neither did conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c nvidia So I visited and followed to implement conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch but it doesn't work. If you want to use the A100-SXM4-40GB GPU with PyTorch, please check the instructions at The current PyTorch install supports CUDA capabilities sm_37 sm_50 sm_60 sm_70. I implemented the totally same command I used for V100 computing cluster, but it doesn't work and I got the error like below /azureml-envs/azureml_9f42dddb00266f3582208ef8cdab4701/lib/python3.7/site-packages/torch/cuda/_init_.py:104: UserWarning:Ī100-SXM4-40GB with CUDA capability sm_80 is not compatible with the current PyTorch installation. I tried to train the model with A100 computing cluster
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |