def upload_debug(self):
"""Uploads the debug log to cloud storage"""
if os.path.exists(self.log_fname):
pusher = FilePusher(self.api)
pusher.update_file("wandb-debug.log", self.log_fname)
pusher.file_changed("wandb-debug.log", self.log_fname)
pusher.finish()
def upload_debug(self):
"""Uploads the debug log to cloud storage"""
if os.path.exists(self.log_fname):
api = InternalApi()
api.set_current_run_id(self.id)
pusher = FilePusher(api)
pusher.update_file("wandb-debug.log", self.log_fname)
pusher.file_changed("wandb-debug.log", self.log_fname)
pusher.finish()
def from_directory(cls,
directory,
project=None,
entity=None,
run_id=None,
api=None,
ignore_globs=None):
api = api or InternalApi()
run_id = run_id or util.generate_id()
run = Run(run_id=run_id, dir=directory)
run_name = None
project_from_meta = None
snap = DirectorySnapshot(directory)
meta = next((p for p in snap.paths if METADATA_FNAME in p), None)
if meta:
meta = json.load(open(meta))
run_name = meta.get("name")
project_from_meta = meta.get("project")
project = project or project_from_meta or api.settings(
"project") or run.auto_project_name(api=api)
if project is None:
raise ValueError("You must specify project")
api.set_current_run_id(run_id)
api.set_setting("project", project)
if entity:
api.set_setting("entity", entity)
res = api.upsert_run(name=run_id,
project=project,
entity=entity,
display_name=run_name)
entity = res["project"]["entity"]["name"]
wandb.termlog("Syncing {} to:".format(directory))
try:
wandb.termlog(res["displayName"] + " " + run.get_url(api))
except CommError as e:
wandb.termwarn(e.message)
file_api = api.get_file_stream_api()
file_api.start()
paths = [
os.path.relpath(abs_path, directory) for abs_path in snap.paths
if os.path.isfile(abs_path)
]
if ignore_globs:
paths = set(paths)
for g in ignore_globs:
paths = paths - set(fnmatch.filter(paths, g))
paths = list(paths)
run_update = {"id": res["id"]}
tfevents = sorted([p for p in snap.paths if ".tfevents." in p])
history = next((p for p in snap.paths if HISTORY_FNAME in p), None)
event = next((p for p in snap.paths if EVENTS_FNAME in p), None)
config = next((p for p in snap.paths if CONFIG_FNAME in p), None)
user_config = next((p for p in snap.paths if USER_CONFIG_FNAME in p),
None)
summary = next((p for p in snap.paths if SUMMARY_FNAME in p), None)
if history:
wandb.termlog("Uploading history metrics")
file_api.stream_file(history)
snap.paths.remove(history)
elif len(tfevents) > 0:
from wandb import tensorflow as wbtf
wandb.termlog("Found tfevents file, converting...")
summary = {}
for path in tfevents:
filename = os.path.basename(path)
namespace = path.replace(filename,
"").replace(directory,
"").strip(os.sep)
summary.update(
wbtf.stream_tfevents(path,
file_api,
run,
namespace=namespace))
for path in glob.glob(os.path.join(directory, "media/**/*"),
recursive=True):
if os.path.isfile(path):
paths.append(path)
else:
wandb.termerror(
"No history or tfevents files found, only syncing files")
if event:
file_api.stream_file(event)
snap.paths.remove(event)
if config:
run_update["config"] = util.load_yaml(open(config))
elif user_config:
# TODO: half backed support for config.json
run_update["config"] = {
k: {
"value": v
}
for k, v in six.iteritems(user_config)
}
if isinstance(summary, dict):
#TODO: summary should already have data_types converted here...
run_update["summary_metrics"] = util.json_dumps_safer(summary)
elif summary:
run_update["summary_metrics"] = open(summary).read()
if meta:
if meta.get("git"):
run_update["commit"] = meta["git"].get("commit")
run_update["repo"] = meta["git"].get("remote")
if meta.get("host"):
run_update["host"] = meta["host"]
run_update["program_path"] = meta["program"]
run_update["job_type"] = meta.get("jobType")
run_update["notes"] = meta.get("notes")
else:
run_update["host"] = run.host
wandb.termlog("Updating run and uploading files")
api.upsert_run(**run_update)
pusher = FilePusher(api)
for k in paths:
path = os.path.abspath(os.path.join(directory, k))
pusher.update_file(k, path)
pusher.file_changed(k, path)
pusher.finish()
pusher.print_status()
file_api.finish(0)
# Remove temporary media images generated from tfevents
if history is None and os.path.exists(os.path.join(directory,
"media")):
shutil.rmtree(os.path.join(directory, "media"))
wandb.termlog("Finished!")
return run
def from_directory(cls, directory, project=None, entity=None, run_id=None, api=None):
api = api or InternalApi()
run_id = run_id or util.generate_id()
run = Run(run_id=run_id, dir=directory)
project = project or api.settings(
"project") or run.auto_project_name(api=api)
if project is None:
raise ValueError("You must specify project")
api.set_current_run_id(run_id)
api.set_setting("project", project)
if entity:
api.set_setting("entity", entity)
res = api.upsert_run(name=run_id, project=project, entity=entity)
entity = res["project"]["entity"]["name"]
wandb.termlog("Syncing {} to:".format(directory))
wandb.termlog(run.get_url(api))
file_api = api.get_file_stream_api()
snap = DirectorySnapshot(directory)
paths = [os.path.relpath(abs_path, directory)
for abs_path in snap.paths if os.path.isfile(abs_path)]
run_update = {"id": res["id"]}
tfevents = sorted([p for p in snap.paths if ".tfevents." in p])
history = next((p for p in snap.paths if HISTORY_FNAME in p), None)
event = next((p for p in snap.paths if EVENTS_FNAME in p), None)
config = next((p for p in snap.paths if CONFIG_FNAME in p), None)
user_config = next(
(p for p in snap.paths if USER_CONFIG_FNAME in p), None)
summary = next((p for p in snap.paths if SUMMARY_FNAME in p), None)
meta = next((p for p in snap.paths if METADATA_FNAME in p), None)
if history:
wandb.termlog("Uploading history metrics")
file_api.stream_file(history)
snap.paths.remove(history)
elif len(tfevents) > 0:
from wandb import tensorflow as wbtf
wandb.termlog("Found tfevents file, converting.")
for file in tfevents:
summary = wbtf.stream_tfevents(file, file_api)
else:
wandb.termerror(
"No history or tfevents files found, only syncing files")
if event:
file_api.stream_file(event)
snap.paths.remove(event)
if config:
run_update["config"] = util.load_yaml(
open(config))
elif user_config:
# TODO: half backed support for config.json
run_update["config"] = {k: {"value": v}
for k, v in six.iteritems(user_config)}
if summary:
run_update["summary_metrics"] = open(summary).read()
if meta:
meta = json.load(open(meta))
if meta.get("git"):
run_update["commit"] = meta["git"].get("commit")
run_update["repo"] = meta["git"].get("remote")
run_update["host"] = meta["host"]
run_update["program_path"] = meta["program"]
run_update["job_type"] = meta.get("jobType")
else:
run_update["host"] = socket.gethostname()
wandb.termlog("Updating run and uploading files")
api.upsert_run(**run_update)
pusher = FilePusher(api)
for k in paths:
path = os.path.abspath(os.path.join(directory, k))
pusher.update_file(k, path)
pusher.file_changed(k, path)
pusher.finish()
pusher.print_status()
wandb.termlog("Finished!")
return run