SimmulatedAnnealingPrune inifinite loop

[duda1202]

Describe the issue:

Environment:

• NNI version: 2.6
• Training service (local|remote|pai|aml|etc):remote
• Client OS: Ubuntu 18
• Server OS (for remote mode only): Ubuntu 18
• Python version: 3.6.9
• PyTorch/TensorFlow version: PyTorch 1.8.1+cu111
• Is conda/virtualenv/venv used?: no
• Is running in Docker?: yes

Configuration:

• Experiment config (remember to remove secrets!):
• Search space:

Log message:

• nnimanager.log:
• dispatcher.log:
• nnictl stdout and stderr:

How to reproduce it?:

I would like to know the function of :

                else:
                    delta_E = np.abs(evaluation_result -
                                     self._current_performance)
                    probability = math.exp(-1 * delta_E /
                                           self._current_temperature)
                    if np.random.uniform(0, 1) < probability:
                        self._current_performance = evaluation_result
                        self._sparsities = sparsities_perturbated
                        break

I get probability values between 1e-11 and 1e-7 and it never breaks from the loop to the next iteration. And I have to restart the script

[duda1202]

I understand now that if you use other metrics that are not between 0-1, the probability values are too low and it will never end the loop. So, if you use RMSE in mm for example, probability values can be too low. Be sure to change it to fit this code.