4.2-paho-mqtt-colab-sdl-demo-test.ipynb error?

[ianfoster]

I'll try to debug this myself, but in case someone knows the answer:

Running in 4.2-paho-mqtt-colab-sdl-demo-test.ipynb on Colab the following simplified code, I get the error shown below ("ValueError: Raw data is expected to be just for one arm.")

from self_driving_lab_demo.utils.search import (
    ax_bayesian_optimization,
)

num_iter = 10

best_parameters, values, experiment, model = ax_bayesian_optimization(sdl, num_iter)

[INFO 01-07 17:24:27] ax.modelbridge.dispatch_utils: Using Bayesian optimization since there are more ordered parameters than there are categories for the unordered categorical parameters.
[INFO 01-07 17:24:27] ax.modelbridge.dispatch_utils: Calculating the number of remaining initialization trials based on num_initialization_trials=None max_initialization_trials=None num_tunable_parameters=3 num_trials=None use_batch_trials=False
[INFO 01-07 17:24:27] ax.modelbridge.dispatch_utils: calculated num_initialization_trials=6
[INFO 01-07 17:24:27] ax.modelbridge.dispatch_utils: num_completed_initialization_trials=0 num_remaining_initialization_trials=6
[INFO 01-07 17:24:27] ax.modelbridge.dispatch_utils: Using Bayesian Optimization generation strategy: GenerationStrategy(name='Sobol+GPEI', steps=[Sobol for 6 trials, GPEI for subsequent trials]). Iterations after 6 will take longer to generate due to model-fitting.
[INFO 01-07 17:24:27] ax.service.managed_loop: Started full optimization with 10 steps.
[INFO 01-07 17:24:27] ax.service.managed_loop: Running optimization trial 1...
[ERROR 01-07 17:24:37] ax.service.managed_loop: Encountered exception during optimization: 
Traceback (most recent call last):
  File "/usr/local/lib/python3.8/dist-packages/ax/service/managed_loop.py", line 227, in full_run
    self.run_trial()
  File "/usr/local/lib/python3.8/dist-packages/ax/utils/common/executils.py", line 161, in actual_wrapper
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.8/dist-packages/ax/service/managed_loop.py", line 204, in run_trial
    _, data = InstantiationBase.data_and_evaluations_from_raw_data(
  File "/usr/local/lib/python3.8/dist-packages/ax/service/utils/instantiation.py", line 970, in data_and_evaluations_from_raw_data
    evaluations = {
  File "/usr/local/lib/python3.8/dist-packages/ax/service/utils/instantiation.py", line 971, in <dictcomp>
    arm_name: cls.raw_data_to_evaluation(
  File "/usr/local/lib/python3.8/dist-packages/ax/service/utils/instantiation.py", line 908, in raw_data_to_evaluation
    raise ValueError("Raw data is expected to be just for one arm.")
ValueError: Raw data is expected to be just for one arm.
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<timed exec> in <module>

[/usr/local/lib/python3.8/dist-packages/self_driving_lab_demo/utils/search.py](https://92s1rnlmebf-496ff2e9c6d22116-0-colab.googleusercontent.com/outputframe.html?vrz=colab-20230105-060053-RC00_499751356#) in ax_bayesian_optimization(sdl, num_iter, objective_name)
     51     parameters = [dict(name=nm, type="range", bounds=bnd) for nm, bnd in bounds.items()]
     52 
---> 53     best_parameters, values, experiment, model = optimize(
     54         parameters=parameters,
     55         evaluation_function=evaluation_function,

3 frames
[/usr/local/lib/python3.8/dist-packages/ax/service/utils/best_point.py](https://92s1rnlmebf-496ff2e9c6d22116-0-colab.googleusercontent.com/outputframe.html?vrz=colab-20230105-060053-RC00_499751356#) in get_best_raw_objective_point_with_trial_index(experiment, optimization_config, trial_indices)
     94     dat = experiment.lookup_data(trial_indices=trial_indices)
     95     if dat.df.empty:
---> 96         raise ValueError("Cannot identify best point if experiment contains no data.")
     97     objective = optimization_config.objective
     98     feasible_df = _filter_feasible_rows(

ValueError: Cannot identify best point if experiment contains no data.

[sgbaird]

Hi @ianfoster, thanks for the detailed info! The fix was on the main branch but not in the latest Python package version. I've updated the Python package to v0.8.0 and verified that it runs on Colab. Lmk if that works for you. For your own device, you may also need to upload the v0.8.0 sdl_demo.zip code to your Pico W.

[ianfoster]

That works, thanks!