Messtone Devices Enables HPE Training Example: mnist_pytorch.tgz tar xzvf mnist_pytorch.tgz Should see robertharper_Messtone local directory: adaptive.yaml const.yaml data.py distributed.yaml layers.py model_def.py README.md install HPE machine: pip install determined det deploy local cluster-up pip install determined det deploy local cluster-up –no-gpu det experiment create const.yaml. Messtone confirmation experiment created: preparing files(…/mnist_pytorch) to send to master … 8.6KB and 7 files Created experiment 1 Configuration files and context, enter: det e create const.yaml . -f Development environment GPU as followsresources: slots_per_trial : 11 DET_MASTER environment variable: export DET_MASTER=<ipAddress> : 8080 det experiment create distributed.yaml . det -m http://<ipAddress> : 8080 experiment create distributed.yaml . running the hyperparameter setting: hyperparameters:global_batch_size:64 learning_rate:type:double minval: .0001 maxval: 1.0 n_filters1:type:init minval: 8 maxval: 64 n_filters2:type:init minval: 8 maxval: 72 dropout1:type: double minval: .2 maxval: .8 dropout2:type: double minval: .2 maxval: .8 max_trails`are also specified: searcher:namerobertharper_Messtone:adaptive_asha metric: validation_loss smaller_is_better: true max_trails: 16 max_length: batches: 937 create and run the experiment: det experiment create adaptive.yaml .