async def delete_gen(request, gen):
if request.args.get('really', 'no') == 'yes':
gen_trials = MongoTrials('mongo://localhost:1234/covid/jobs',
f'covid-{gen}')
gen_trials.refresh()
gen_trials.delete_all()
return redirect(f"/status/?refresh=true")
return html(create_delete_prompt(f"GENERATION 'covid-{gen}'"))
def _refresh_trials():
global TRIALS, TRIALS_REFRESHED
if TRIALS is None:
TRIALS = MongoTrials('mongo://localhost:1234/covid/jobs')
TRIALS_REFRESHED = datetime.now()
TRIALS.refresh()
def run_optimization(level=1):
print(f"Optimizing at level {level}")
set_random_seeds(4)
next_lvl_trials = MongoTrials('mongo://localhost:1234/covid/jobs',
exp_key=f'covid-{level+1}')
if len(next_lvl_trials.trials) > 0:
print(f"Already completed level {level} -- skipping")
return
exp_key = f'covid-{level}'
trials = MongoTrials('mongo://localhost:1234/covid/jobs', exp_key=exp_key)
suggestion_box = hyperopt.tpe.suggest
if level == 1:
max_evals = LEVEL_DEFS[0][1]
depth = 1
elif level > 1:
depth, new_budget, extend_budget = LEVEL_DEFS[level - 1]
last_depth, _, _ = LEVEL_DEFS[level - 2]
# Minimum one per node for the expensive ones -- no point wasting compute time
num_new = int(np.ceil((new_budget / depth) / NUM_NODES) * NUM_NODES)
if len(trials.trials) == 0:
print("Generating estimates from previous level")
result_docs = configure_next_level(level, depth, extend_budget)
num_to_extend = len(result_docs)
suggestion_box = create_suggestion_box(result_docs)
last_level_trials = MongoTrials('mongo://localhost:1234/covid/jobs',
exp_key=f'covid-{level-1}')
prev_level_count = len(
[x for x in last_level_trials.losses() if x is not None])
max_evals = prev_level_count + num_new
trials.refresh()
objective = functools.partial(test_parameterization, num_epochs=depth)
if len([x for x in trials.statuses() if x == 'ok']) >= max_evals:
print(f"Already completed level {level} -- skipping")
else:
best = hyperopt.fmin(objective,
space=SEARCH_SPACE,
algo=suggestion_box,
max_evals=max_evals,
trials=trials)
print(best)
def validate_from_tids(dbname):
trials = MongoTrials('mongo://localhost:44556/%s/jobs' % dbname,
refresh=False)
trials.refresh()
tdict = dict([(t['tid'], t) for t in trials])
print "TIDS", tdict.keys()
for tid, t in tdict.items():
assert t['misc']['tid'] == tid
if 'from_tid' in t['misc']:
if t['misc']['from_tid'] not in tdict:
print 'WTF gave us', tid, t['misc']['from_tid']
def main():
#logging.basicConfig(level=logging.INFO)
data_root = abspath2("../reference-data")
input_root = abspath2("../bayesian-input")
output_root = abspath2("../bayesian-output")
os.makedirs(input_root, exist_ok=True)
os.makedirs(output_root, exist_ok=True)
seed = 123
np.random.seed(seed)
# Global variable
global ITERATION
ITERATION = 0
MAX_EVALS = 2
from hyperopt import fmin
from hyperopt import tpe
portion = 0.5
budgetfile = "budget.csv"
if not os.path.exists('/home/ubuntu/BeamCompetitions/budget.csv'):
data = pd.DataFrame(columns=['iter'])
print(data)
data.to_csv('/home/ubuntu/BeamCompetitions/budget.csv')
#initialize the 1st mongotrail
bayes_trials = MongoTrials('mongo://localhost:27017/wh_db/jobs',
exp_key=str(datetime.datetime.now()))
len_trial = 21
while len_trial >= 20:
#bayes_trials = MongoTrials('mongo://localhost:27017/wh_db/jobs', exp_key=str(datetime.datetime.now()))
#print("After getting trials, I can print")
best = fmin(fn=objective,
space=space,
algo=tpe.suggest,
max_evals=MAX_EVALS,
trials=bayes_trials,
rstate=np.random.RandomState(50))
budget = pd.read_csv("/home/ubuntu/BeamCompetitions/budget.csv")
if (len(budget) == 0): n_sim_iters = 1
else:
n_sim_iters = budget.loc[len(budget) - 1, 'iter'] + budget_update
n_sim_iters = int(n_sim_iters)
#budget.loc[len(budget),'iter']=n_sim_iters
budget = budget.append({'iter': n_sim_iters}, ignore_index=True)
budget.to_csv("/home/ubuntu/BeamCompetitions/budget.csv", index=False)
sorted_trial = sorted(list(bayes_trials.trials),
key=lambda x: x['result']['loss'])
list_trial = list(sorted_trial)
cutted_trial = list_trial[0:math.floor(len(list_trial) * portion)]
MAX_EVALS = math.floor(len(list_trial) * portion)
#suppose the cutted_trial is correct
new_exp = str(datetime.datetime.now())
test_trials = MongoTrials('mongo://localhost:27017/wh_db/jobs',
exp_key=new_exp)
for trial in cutted_trial:
hyperopt_trial = hyperopt.Trials(new_exp).new_trial_docs(
tids=[trial['tid']],
specs=[trial['spec']],
results=[trial['result']],
miscs=[trial['misc']])
hyperopt_trial[0]['state'] = hyperopt.JOB_STATE_DONE
test_trials.insert_trial_docs(hyperopt_trial)
test_trials.refresh()
bayes_trials = test_trials
len_trial = len(list(bayes_trials.trials))
if not os.path.exists('../bayesian-output/bayes_trials.csv'):
data = pd.DataFrame(
columns=['loss', 'params', 'iteration', 'train_time'])
data.to_csv('../bayesian-output/bayes_trials.csv')
outputfile = pd.read_csv("../bayesian-output/bayes_trials.csv")
for result in list(bayes_trials.results):
#of_connection.write(",".join(result))
outputfile = outputfile.append(
{
'loss': result['loss'],
'params': list(result['params']),
'iteration': result['iter'],
'train_time': result['train_time']
},
ignore_index=True)
# trials = self.convertResultsToTrials(hyperparameterSpace, filteredResults),
# iter=iter+1
outputfile.to_csv('../bayesian-output/bayes_trials.csv')
