Which AutoML method is used in the current version of PocketFlow？

Tencent / PocketFlow

An Automatic Model Compression (AutoMC) framework for developing smaller and faster AI applications.

https://pocketflow.github.io

Other

2.79k stars 490 forks source link

Which AutoML method is used in the current version of PocketFlow？ #149

Closed Moran232 closed 5 years ago

Moran232 commented 5 years ago

Hi, you mentioned in Github that Gaussian Processes (GP, Mockus, 1975), Tree-structured Parzen Estimator (TPE, Bergstra et al., 2013), and Deterministic Deep Policy Gradients (DDPG, Lillicrap et al., 2016) will be used for hyper-parameters optimization.

While it seems that the automl methods GP and TPE haven't been used in the current version, right?(Only RL-DDPG) Will you open the source code in recent days?

jiaxiang-wu commented 5 years ago

Yes, both GP and TPE are not included in the current repo. They are based on another project which is not released. We are planning to provide GP and TPE support within PocketFlow, but there is no clear timetable for this, since our current priority is to support object detection models.

aakejiang commented 1 year ago

@jiaxiang-wu How about now, are GP and TPE supported? Since there are still no implementations in "automl" module.