Daniel Ho

Thanks for checking. The augmentations applied in `pba.ipynb` are explicitly shuffled. Do you have a line number you are referring to? I don't believe any dict is being read in...

Ah, thanks for the great catch! I've reordered the dict to the python 2 ordering which I believe stays consistent.

Hi, thanks for trying to reproduce! This is much lower than expected, and likely because of the all 0 schedule you're seeing (this effectively does not use any PBA augmentation)....

Apologies for the late reply. I tried to use your setup but I couldn't reproduce your issue where you run into a final schedule of all 0s. Perhaps you can...

If you want to just reproduce the results of the search without running it yourself, you can use the preexisting schedules at https://github.com/arcelien/pba/tree/master/schedules. These are automatically loaded when you run...

This would be dependent on Ray's PBT implementation's support of mutli-gpu. When I tried in Feb/March of this year, there were some issues which may have been resolved.

Sorry, that scrupt actually reflects the main results in table 2, not the ablation in table 3. I will rename the scripts which were named according to an old version...

Hey, I unfortunately haven't seen that error before. Perhaps you can rerun the code to see if it's just an intermittent issue, or open an issue over at https://github.com/ray-project/ray/issues.

Unfortunately I have not tried this yet for object detection. You would need to change the set of augmentations operations to take bounding boxes as input, something like https://arxiv.org/abs/1906.11172.

Although the code has only been tested with 1 GPU, you usually have to specify CUDA_VISIBLE_DEVICES to constrain the search. Perhaps a newer version of Ray makes this constraint instead?