Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix and refactor parallelize_sessions #174

Merged
merged 1 commit into from
Sep 21, 2018
Merged

fix and refactor parallelize_sessions #174

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
40 changes: 18 additions & 22 deletions slm_lab/experiment/control.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -173,25 +173,29 @@ def __init__(self, spec, info_space):
self.mp_runner = init_run_session if self.is_singleton else init_run_space_session
logger.info(f'Initialized trial {self.index}')

def init_session_and_run(self, info_space):
session = self.SessionClass(self.spec, info_space)
session_data = session.run()
return session_data

def run_sessions(self):
logger.info('Running sessions')
info_spaces = []
def parallelize_sessions(self, global_nets=None):
workers = []
for _s in range(self.spec['meta']['max_session']):
self.info_space.tick('session')
info_spaces.append(deepcopy(self.info_space))
w = mp.Process(target=self.mp_runner, args=(deepcopy(self.spec), self.info_space, global_nets))
w.start()
workers.append(w)
for w in workers:
w.join()
session_datas = analysis.session_data_dict_for_dist(self.spec, self.info_space)
return session_datas

if util.get_lab_mode() == 'train' and len(info_spaces) > 1:
def run_sessions(self):
logger.info('Running sessions')
if util.get_lab_mode() == 'train' and self.spec['meta']['max_session'] > 1:
# when training a single spec over multiple sessions
session_datas = util.parallelize_fn(self.init_session_and_run, info_spaces, ps.get(self.spec['meta'], 'resources.num_cpus', util.NUM_CPUS))
session_datas = self.parallelize_sessions()
else:
session_datas = []
for info_space in info_spaces:
session_data = self.init_session_and_run(info_space)
for _s in range(self.spec['meta']['max_session']):
self.info_space.tick('session')
session = self.SessionClass(self.spec, self.info_space)
session_data = session.run()
session_datas.append(session_data)
if analysis.is_unfit(session_data):
break
Expand Down Expand Up @@ -219,15 +223,7 @@ def init_global_nets(self):
def run_distributed_sessions(self):
logger.info('Running distributed sessions')
global_nets = self.init_global_nets()
workers = []
for _s in range(self.spec['meta']['max_session']):
self.info_space.tick('session')
w = mp.Process(target=self.mp_runner, args=(deepcopy(self.spec), self.info_space, global_nets))
w.start()
workers.append(w)
for w in workers:
w.join()
session_datas = analysis.session_data_dict_for_dist(self.spec, self.info_space)
session_datas = self.parallelize_sessions(global_nets)
return session_datas

def close(self):
Expand Down