def __init__(self,
hyperopt_sampler,
output_feature: str,
metric: str,
split: str,
cpu_resources_per_trial: int = None,
gpu_resources_per_trial: int = None,
kubernetes_namespace: str = None,
**kwargs) -> None:
if ray is None:
raise ImportError('ray module is not installed. To '
'install it,try running pip install ray')
if not isinstance(hyperopt_sampler, RayTuneSampler):
raise ValueError(
'Sampler {} is not compatible with RayTuneExecutor, '
'please use the RayTuneSampler'.format(hyperopt_sampler))
HyperoptExecutor.__init__(self, hyperopt_sampler, output_feature,
metric, split)
try:
ray.init('auto', ignore_reinit_error=True)
except ConnectionError:
logger.info('Initializing new Ray cluster...')
ray.init()
self.search_space = hyperopt_sampler.search_space
self.num_samples = hyperopt_sampler.num_samples
self.goal = hyperopt_sampler.goal
self.search_alg_dict = hyperopt_sampler.search_alg_dict
self.output_feature = output_feature
self.metric = metric
self.split = split
self.trial_id = 0
self.cpu_resources_per_trial = cpu_resources_per_trial
self.gpu_resources_per_trial = gpu_resources_per_trial
self.kubernetes_namespace = kubernetes_namespace
def get_metric_score(self, train_stats, eval_stats) -> float:
if self._has_metric(train_stats, TEST):
logger.info(
"Returning metric score from training (test) statistics")
return self.get_metric_score_from_train_stats(train_stats, TEST)
elif self._has_eval_metric(eval_stats):
logger.info("Returning metric score from eval statistics. "
"If skip_save_model is True, eval statistics "
"are calculated using the model at the last epoch "
"rather than the model at the epoch with "
"best validation performance")
return self.get_metric_score_from_eval_stats(eval_stats)
elif self._has_metric(train_stats, VALIDATION):
logger.info(
"Returning metric score from training (validation) statistics")
return self.get_metric_score_from_train_stats(
train_stats, VALIDATION)
elif self._has_metric(train_stats, TRAINING):
logger.info(
"Returning metric score from training split statistics, "
"as no test / validation / eval sets were given")
return self.get_metric_score_from_train_stats(
train_stats, TRAINING)
else:
raise RuntimeError(
"Unable to obtain metric score from missing training / eval statistics"
)
def get_metric_score(self, train_stats, eval_stats) -> float:
if (train_stats is not None and self.split in train_stats
and VALIDATION in train_stats
and # needed otherwise can-t figure
# out the best epoch
self.output_feature in train_stats[self.split] and self.metric
in train_stats[self.split][self.output_feature]):
logger.info("Returning metric score from training statistics")
return self.get_metric_score_from_train_stats(train_stats)
else:
logger.info("Returning metric score from eval statistics. "
"If skip_save_model is True, eval statistics "
"are calculated using the model at the last epoch "
"rather than the model at the epoch with "
"best validation performance")
return self.get_metric_score_from_eval_stats(eval_stats)
def __init__(
self,
hyperopt_sampler,
output_feature: str,
metric: str,
split: str,
cpu_resources_per_trial: int = None,
gpu_resources_per_trial: int = None,
kubernetes_namespace: str = None,
time_budget_s: Union[int, float, datetime.timedelta] = None,
max_concurrent_trials: Optional[int] = None,
**kwargs,
) -> None:
if ray is None:
raise ImportError("ray module is not installed. To "
"install it,try running pip install ray")
if not isinstance(hyperopt_sampler, RayTuneSampler):
raise ValueError(
"Sampler {} is not compatible with RayTuneExecutor, "
"please use the RayTuneSampler".format(hyperopt_sampler))
HyperoptExecutor.__init__(self, hyperopt_sampler, output_feature,
metric, split)
if not ray.is_initialized():
try:
ray.init("auto", ignore_reinit_error=True)
except ConnectionError:
logger.info("Initializing new Ray cluster...")
ray.init(ignore_reinit_error=True)
self.search_space = hyperopt_sampler.search_space
self.num_samples = hyperopt_sampler.num_samples
self.goal = hyperopt_sampler.goal
self.search_alg_dict = hyperopt_sampler.search_alg_dict
self.scheduler = hyperopt_sampler.scheduler
self.decode_ctx = hyperopt_sampler.decode_ctx
self.output_feature = output_feature
self.metric = metric
self.split = split
self.trial_id = 0
self.cpu_resources_per_trial = cpu_resources_per_trial
self.gpu_resources_per_trial = gpu_resources_per_trial
self.kubernetes_namespace = kubernetes_namespace
self.time_budget_s = time_budget_s
self.max_concurrent_trials = max_concurrent_trials
self.sync_config = None