Getting `UnparsedFlagAccessError` with example code

[Jeremy Morgan]

Hi there,

I'm trying to run the demo code from the README, but am getting an UnparsedFlagAccessError error.

Code:

from vizier.service import clients
from vizier.service import pyvizier as vz

# Objective function to maximize.
def evaluate(w: float, x: int, y: float, z: str) -> float:
    return w**2 - y**2 + x * ord(z)

# Algorithm, search space, and metrics.
study_config = vz.StudyConfig(algorithm='GAUSSIAN_PROCESS_BANDIT')
study_config.search_space.root.add_float_param('w', 0.0, 5.0)
study_config.search_space.root.add_int_param('x', -2, 2)
study_config.search_space.root.add_discrete_param('y', [0.3, 7.2])
study_config.search_space.root.add_categorical_param('z', ['a', 'g', 'k'])
study_config.metric_information.append(vz.MetricInformation('metric_name', goal=vz.ObjectiveMetricGoal.MAXIMIZE))

# Setup client and begin optimization. Vizier Service will be implicitly created.
study = clients.Study.from_study_config(study_config, owner='my_name', study_id='example')
for i in range(10):
    suggestions = study.suggest(count=1)
    for suggestion in suggestions:
        params = suggestion.parameters
        objective = evaluate(params['w'], params['x'], params['y'], params['z'])
        suggestion.complete(vz.Measurement({'metric_name': objective}))

Error:

---------------------------------------------------------------------------
UnparsedFlagAccessError                   Traceback (most recent call last)
Cell In[21], line 20
     18 study = clients.Study.from_study_config(study_config, owner='my_name', study_id='example')
     19 for i in range(10):
---> 20     suggestions = study.suggest(count=1)
     21     for suggestion in suggestions:
     22         params = suggestion.parameters

File ~/Projects/cppflow/venv/lib/python3.8/site-packages/vizier/_src/service/clients.py:143, in Study.suggest(self, count, client_id)
    138 def suggest(
    139     self, *, count: Optional[int] = None, client_id: str = 'default_client_id'
    140 ) -> List[Trial]:
    141   return [
    142       self._trial_client(t)
--> 143       for t in self._client.get_suggestions(
    144           count, client_id_override=client_id
    145       )
    146   ]

File ~/Projects/cppflow/venv/lib/python3.8/site-packages/vizier/_src/service/vizier_client.py:168, in VizierClient.get_suggestions(self, suggestion_count, client_id_override)
    165 num_attempts = 0
    166 while not operation.done:
    167   sleep_time = PollingDelay(
--> 168       num_attempts, FLAGS.vizier_new_suggestion_polling_secs
    169   )
    170   num_attempts += 1
    171   logging.info(
    172       'Waiting for operation with name %s to be done', operation.name
    173   )

File ~/Projects/cppflow/venv/lib/python3.8/site-packages/absl/flags/_flagvalues.py:481, in FlagValues.__getattr__(self, name)
    479   return fl[name].value
    480 else:
--> 481   raise _exceptions.UnparsedFlagAccessError(
    482       'Trying to access flag --%s before flags were parsed.' % name)

UnparsedFlagAccessError: Trying to access flag --vizier_new_suggestion_polling_secs before flags were parsed.

Specs:

• Ubuntu 20, python3.8, google-vizier=0.1.5

Any ideas on how to fix this? I looked around online but didn't find anything helpful.

Thanks,
Jeremy

[Anton Lee]

The project has some hidden CLI flags that need to be parsed, setting them to a default value.
Adding the following at the top of my script worked for me.

from absl.flags import FLAGS
import sys
FLAGS(sys.argv)

Would be good to see the project either document the step or make it redundant.

[Jeremy Morgan]

Thanks it's working now. I was getting an error UnrecognizedFlagError: Unknown command line flag 'f' when running FLAGS(sys.argv). My sys.argv is ['/home/.../venv/lib/python3.8/site-packages/ipykernel_launcher.py', '-f', '/home/.../.local/share/jupyter/runtime/kernel-ead04449-484c-47c8-9f40-a6ded4d1e08a.json']. I changed the code to the following and it worked:

from absl.flags import FLAGS
import sys
FLAGS([sys.argv[0]])

Should this change by submitted as a PR to the readme? happy to do so

[Jeremy Morgan]

Another thing, this call suggestions = study.suggest(count=1) is hanging. Any idea why that is? Do I need to update the client_id parameter or something?

thanks

[Anton Lee]

I recall running into this as well. It fixed itself when I followed the read-the-docs documentation rather than the readme https://oss-vizier.readthedocs.io/en/latest/guides/user/running_vizier.html. Sorry, I can't remember what exactly fixed it.

[Anton Lee]

This code works for me

from vizier.service import clients
from vizier.service import pyvizier as vz


from absl.flags import FLAGS

FLAGS([""])

problem = vz.ProblemStatement()
problem.search_space.root.add_float_param("x", 0.0, 1.0)
problem.search_space.root.add_float_param("y", 0.0, 1.0)
problem.metric_information.append(
    vz.MetricInformation(name="maximize_metric", goal=vz.ObjectiveMetricGoal.MAXIMIZE)
)


def evaluate(x: float, y: float) -> float:
    return x**2 - y**2


study_config = vz.StudyConfig.from_problem(problem)
study_config.algorithm = "QUASI_RANDOM_SEARCH"


study_client = clients.Study.from_study_config(
    study_config, owner="owner", study_id="example_study_id"
)


for i in range(10):
    suggestions = study_client.suggest(count=1)
    for suggestion in suggestions:
        x = suggestion.parameters["x"]
        y = suggestion.parameters["y"]
        objective = evaluate(x, y)
        print(
            f"Iteration {i}, suggestion ({x},{y}) led to objective value {objective}."
        )
        final_measurement = vz.Measurement({"maximize_metric": objective})
        suggestion.complete(final_measurement)

[xingyousong]

Hi, this should be fixed now in v0.1.6. Thanks!