def detail(training_task_code):
company_id = g.user.company_id
training_task_entity = TrainingTaskEntity.get_for_company_id(
company_id).filter(TrainingTaskEntity.task_code == training_task_code).one_or_none()
if not training_task_entity:
return handle_error(404, 'No training task found!')
context = {
'training_task': TrainingTask.from_model(training_task_entity)
}
response = ApiResponse(
content_type=request.accept_mimetypes.best,
context=context,
template='training/detail.html'
)
return response()
def get_for_task_code(task_code):
model = TrainingTaskEntity.get_by_task_code(task_code)
return TrainingTask.from_model(model)
def submit():
user_id = g.user.id
company_id = g.user.company_id
name = g.json.get('name')
enable_fft = g.json.get('enable_fft', None)
downsample_factor = int(g.json.get('downsample_factor'))
train_iters = int(g.json.get('train_iters'))
if enable_fft:
enable_fft = bool(int(enable_fft))
datasource_configuration_id = g.json.get('datasource_configuration_id')
parent_training_id = g.json.get('parent_training_id', None)
if TrainingTaskEntity.query.filter(TrainingTaskEntity.name == name,
TrainingTaskEntity.company_id == company_id).all():
return handle_error(400, f'Training with name {name} already exists')
latest_company_configuration = g.user.company.current_configuration
datasource_configuration = DataSourceConfigurationEntity.query.filter(
DataSourceConfigurationEntity.company_id == company_id
).filter(
DataSourceConfigurationEntity.id == datasource_configuration_id
).one_or_none()
if not datasource_configuration:
return handle_error(404, 'No datasource type found!')
number_of_sensors = datasource_configuration.meta.get('number_of_sensors')
if not number_of_sensors:
return handle_error(400, 'No number of sensors specified! Check datasource configuration')
number_of_timesteps = 784 // number_of_sensors * downsample_factor
datasources_for_train = DataSourceEntity.get_for_datasource_configuration(datasource_configuration)
if not datasources_for_train:
return handle_error(400, f'No valid datasources available for type {datasource_configuration.name}')
training_task_code = services.detection.generate_task_code()
training_configuration = services.training.create_training_configuration(
training_task_code,
company_id,
latest_company_configuration.configuration,
datasource_configuration.meta,
enable_fft,
parent_training_id
)
updated_training_configuration = add_training_parameters_to_configuration(
downsample_factor, number_of_timesteps, train_iters, training_configuration
)
logging.info(f"created training configuration: {updated_training_configuration}")
training_task_entity = TrainingTaskEntity(
company_id=company_id,
user_id=user_id,
task_code=training_task_code,
company_configuration_id=latest_company_configuration.id,
datasource_configuration_id=datasource_configuration_id,
datasources=datasources_for_train,
configuration=updated_training_configuration,
name=name,
parent_training_id=int(parent_training_id) if parent_training_id else None)
training_task_entity.save()
os.makedirs(training_task_entity.train_data_dir, exist_ok=True)
services.training.set_task_status(training_task_entity, status=TaskStatusTypes.queued, message='Enqueued')
train_celery_task.apply_async(
args=(training_task_entity.task_code,), task_id=training_task_code
)
response = ApiResponse(
content_type=request.accept_mimetypes.best,
context=TrainingTask.from_model(training_task_entity),
status_code=201,
next=url_for('training.list')
)
return response()
def insert(traing_task):
model = traing_task.to_model()
model.save()
return TrainingTask.from_model(model)