def retrieve_summary(model_task='.*',
model_name='.*',
dataset_name='.*',
expname='.*',
score_name='.*',
split='.*',
model_id=None):
"""
Retrieve previously computed summary
"""
with session_scope() as sess:
query = sess.query(VegaiModelzoo).\
filter(VegaiModelzoo.model_name.op('~')(model_name)).\
filter(VegaiModelzoo.dataset_name.op('~')(dataset_name)).\
filter(VegaiModelzoo.model_task.op('~')(model_task)).\
filter(VegaiModelzoo.score_name.op('~')(score_name)).\
filter(VegaiModelzoo.split.op('~')(split)).\
filter(VegaiModelzoo.expname.op('~')(expname))
if model_id is not None:
query = query.filter(VegaiModelzoo.model_id == model_id)
records = query.all()
entries = []
for record in records:
record = Box(record.__dict__)
record.pop('_sa_instance_state')
entries.append(record)
return pd.DataFrame(entries)
def update_task(self, expname, **kwargs):
with session_scope() as sess:
entry = dict(model_task=self.model_task,
model_name=self.model_name,
expname=expname)
model_id = hashlib.sha1(json.dumps(entry).encode()).hexdigest()
entry = sess.query(VegaiModelzoo).filter(
VegaiModelzoo.model_id == model_id).all()[0]
for key, value in kwargs.items():
setattr(entry, key, value)
def log_task(self, expname: str):
with session_scope() as sess:
entry = dict(model_task=self.model_task,
model_name=self.model_name,
expname=expname)
model_id = self.get_model_id(expname)
entry = VegaiModelzoo(
model_id=model_id,
status="started",
created_on=datetime.datetime.now().strftime("%Y-%m-%d %H:%M"),
**entry)
sess.add(entry)
def get_dataset_name():
with dionysos.session_scope() as sess:
names = sess.query(dionysos.Dataset.dataset_name).all()
names = [
n.dataset_name for n in names
if bool(re.match('cloglossv0.*(?<!overfit)$', n.dataset_name))
]
i = 0
while True:
proposal = 'cloglossv0d{}'.format(i)
if proposal not in names:
break
i += 1
return proposal
def next_trial_name(self):
with session_scope() as sess:
results = sess.query(VegaiModelzoo).filter(
VegaiModelzoo.model_task == self.model_task).filter(
VegaiModelzoo.model_name == self.model_name).filter(
VegaiModelzoo.expname.op('~')("{}.*".format(
self.config_name))).all()
expnames = [r.expname for r in results]
if not expnames:
idx = 0
else:
idx = max(
[int(expname.split('t')[-1]) for expname in expnames])
expname = "{}t{}".format(self.config_name, idx + 1)
return expname
def add_entry_to_dataset_table(entry):
with session_scope() as sess:
sess.query(Dataset).filter(
Dataset.dataset_name == entry.dataset_name).delete()
sess.add(entry)