pyNNsMD.src package

Submodules

pyNNsMD.src.device module

Sets the devices for training scripts.

pyNNsMD.src.device.set_gpu(gpu_ids_list)

Set the visible devices from a list of GPUs. Used to assign a process to a separate GPU.

Also very important is to restrict memeory growth since a single tensorfow process will allocate almost all GPU memory, so two fits can not run on same GPU.

Parameters

gpu_ids_list (list) – Device list.

Returns

None.

pyNNsMD.src.fit module

pyNNsMD.src.fit.fit_model_by_script(i, fit_script, g, filepath, m, proc_async)

Run the training script in subprocess.

Parameters

• i (int) – Index of model.

• fit_script (str) – Name of the training routine.

• filepath (str) – Filepath to model.

• g (int) – GPU index to use.

• m (str) – Fitmode.

• proc_async (bool) –

Returns

subprocess

pyNNsMD.src.fit.fit_model_get_python_cmd_os()

Return proper commandline command for pyhton depending on os.

Returns

Python command either python or pyhton3.

Return type

str

pyNNsMD.src.fit.get_path_for_fit_script(fit_script)

pyNNsMD.src.selection module

Module contents