wassname/ray
1
0
Fork 0
mirror of https://github.com/wassname/ray.git synced 2026-08-10 12:30:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

[tune] use dated experiment dir per default (#11104)

Browse Source
This commit is contained in:
Kai Fricke
2020-09-30 14:43:59 -07:00
committed by GitHub
parent f0dba6bd2b
commit c77cfaa5ad
9 changed files with 82 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/ray/tune/experiment.py
+23 -5
View File
@@ -10,7 +10,7 @@ from ray.tune.result import DEFAULT_RESULTS_DIR
from ray.tune.sample import Domain
from ray.tune.stopper import CombinedStopper, FunctionStopper, Stopper, \
TimeoutStopper
from ray.tune.utils import detect_checkpoint_function
from ray.tune.utils import date_str, detect_checkpoint_function
logger = logging.getLogger(__name__)
@@ -137,8 +137,18 @@ class Experiment:
"within your trainable function.")
self._run_identifier = Experiment.register_if_needed(run)
self.name = name or self._run_identifier
# If the name has been set explicitly, we don't want to create
# dated directories. The same is true for string run identifiers.
if int(os.environ.get("TUNE_DISABLE_DATED_SUBDIR", 0)) == 1 or name \
or isinstance(run, str):
self.dir_name = self.name
else:
self.dir_name = "{}_{}".format(self.name, date_str())
if upload_dir:
self.remote_checkpoint_dir = os.path.join(upload_dir, self.name)
self.remote_checkpoint_dir = os.path.join(upload_dir,
self.dir_name)
else:
self.remote_checkpoint_dir = None
@@ -249,8 +259,16 @@ class Experiment:
return run_object
elif isinstance(run_object, type) or callable(run_object):
name = "DEFAULT"
if hasattr(run_object, "__name__"):
name = run_object.__name__
if hasattr(run_object, "_name"):
name = run_object._name
elif hasattr(run_object, "__name__"):
fn_name = run_object.__name__
if fn_name == "<lambda>":
name = "lambda"
elif fn_name.startswith("<"):
name = "DEFAULT"
else:
name = fn_name
else:
logger.warning(
"No name detected on trainable. Using {}.".format(name))
@@ -287,7 +305,7 @@ class Experiment:
@property
def checkpoint_dir(self):
if self.local_dir:
return os.path.join(self.local_dir, self.name)
return os.path.join(self.local_dir, self.dir_name)
@property
def run_identifier(self):