codes.train package#
Submodules#
codes.train.train_fcts module#
- codes.train.train_fcts.create_task_list_for_surrogate(config, surr_name)#
Creates a list of training tasks for a specific surrogate model based on the configuration file.
- Parameters:
config (dict) – The configuration dictionary taken from the config file.
surr_name (str) – The name of the surrogate model.
- Returns:
A list of training tasks for the surrogate model.
- Return type:
list
- codes.train.train_fcts.parallel_training(tasks, device_list, task_list_filepath)#
Execute the queued training tasks across multiple devices using worker threads.
- Parameters:
tasks (list[tuple]) – Output of
create_task_list_for_surrogate().device_list (list[str]) – Devices allocated to training (e.g. [“cuda:0”, “cuda:1”]).
task_list_filepath (str) – Path to the persisted JSON task list that tracks progress.
- Returns:
Elapsed wall-clock time reported by the shared progress bar.
- Return type:
float
- codes.train.train_fcts.sequential_training(tasks, device_list, task_list_filepath)#
Run all training tasks sequentially on a single device.
- Parameters:
tasks (list[tuple]) – Task specification tuples generated from the config.
device_list (list[str]) – Contains exactly one element (typically “cpu” or a single CUDA id).
task_list_filepath (str) – Path to the JSON file used to resume interrupted runs.
- Returns:
Total elapsed time once all tasks finish.
- Return type:
float
- codes.train.train_fcts.train_and_save_model(surr_name, mode, metric, training_id, seed=None, epochs=None, device='cpu', position=1, threadlock=<codes.train.train_fcts.DummyLock object>)#
Train and save a model for a specific benchmark mode.
- Parameters:
surr_name (str) – The name of the surrogate model.
mode (str) – The benchmark mode.
metric (int) – The metric for the benchmark mode.
training_id (str) – The training ID for the current training session.
seed (int, optional) – Random seed for training.
epochs (int, optional) – Number of training epochs.
device (str, optional) – Device to run training on.
position (int, optional) – Model position in the task list.
threadlock (threading.Lock, optional) – Lock for deterministic threading.
- codes.train.train_fcts.worker(task_queue, device, device_idx, overall_progress_bar, task_list_filepath, errors_encountered, threadlock)#
Worker function to process tasks from the task queue on the given device.
- Parameters:
task_queue (Queue) – The in-memory queue containing the training tasks.
device (str) – The device to use for training.
device_idx (int) – The index of the device in the device list.
overall_progress_bar (tqdm) – The overall progress bar for the training.
task_list_filepath (str) – The filepath to the JSON task list.
errors_encountered (list[bool]) – A shared mutable flag array indicating if an error has occurred (True if at least one task failed).
threadlock (threading.Lock) – A lock to prevent threading issues with PyTorch.
Module contents#
- codes.train.create_task_list_for_surrogate(config, surr_name)#
Creates a list of training tasks for a specific surrogate model based on the configuration file.
- Parameters:
config (dict) – The configuration dictionary taken from the config file.
surr_name (str) – The name of the surrogate model.
- Returns:
A list of training tasks for the surrogate model.
- Return type:
list
- codes.train.parallel_training(tasks, device_list, task_list_filepath)#
Execute the queued training tasks across multiple devices using worker threads.
- Parameters:
tasks (list[tuple]) – Output of
create_task_list_for_surrogate().device_list (list[str]) – Devices allocated to training (e.g. [“cuda:0”, “cuda:1”]).
task_list_filepath (str) – Path to the persisted JSON task list that tracks progress.
- Returns:
Elapsed wall-clock time reported by the shared progress bar.
- Return type:
float
- codes.train.sequential_training(tasks, device_list, task_list_filepath)#
Run all training tasks sequentially on a single device.
- Parameters:
tasks (list[tuple]) – Task specification tuples generated from the config.
device_list (list[str]) – Contains exactly one element (typically “cpu” or a single CUDA id).
task_list_filepath (str) – Path to the JSON file used to resume interrupted runs.
- Returns:
Total elapsed time once all tasks finish.
- Return type:
float
- codes.train.train_and_save_model(surr_name, mode, metric, training_id, seed=None, epochs=None, device='cpu', position=1, threadlock=<codes.train.train_fcts.DummyLock object>)#
Train and save a model for a specific benchmark mode.
- Parameters:
surr_name (str) – The name of the surrogate model.
mode (str) – The benchmark mode.
metric (int) – The metric for the benchmark mode.
training_id (str) – The training ID for the current training session.
seed (int, optional) – Random seed for training.
epochs (int, optional) – Number of training epochs.
device (str, optional) – Device to run training on.
position (int, optional) – Model position in the task list.
threadlock (threading.Lock, optional) – Lock for deterministic threading.
- codes.train.worker(task_queue, device, device_idx, overall_progress_bar, task_list_filepath, errors_encountered, threadlock)#
Worker function to process tasks from the task queue on the given device.
- Parameters:
task_queue (Queue) – The in-memory queue containing the training tasks.
device (str) – The device to use for training.
device_idx (int) – The index of the device in the device list.
overall_progress_bar (tqdm) – The overall progress bar for the training.
task_list_filepath (str) – The filepath to the JSON task list.
errors_encountered (list[bool]) – A shared mutable flag array indicating if an error has occurred (True if at least one task failed).
threadlock (threading.Lock) – A lock to prevent threading issues with PyTorch.