@matthewachan Hi, thank you for sharing the code. I am having trouble reproducing the results on the toy dataset. Note: I also had to modify the Makefile to run the toy baseline, but these were fairly minor changes (see my fork).
Below are the HyperDM results, in which it appears far more uncertain in all regions compared to the picture in the README file.
Below are the ensemble results, which roughly resemble the picture in the README but with a noticeably larger uncertainty region.
I was wondering if the training/evaluation parameters were perhaps incorrect and led to this discrepancy. The code for reproducing these results is available on my fork, I made minimal changes just to run the program since there were some errors initially.
@matthewachan Hi, thank you for sharing the code. I am having trouble reproducing the results on the toy dataset. Note: I also had to modify the Makefile to run the toy baseline, but these were fairly minor changes (see my fork).
Below are the HyperDM results, in which it appears far more uncertain in all regions compared to the picture in the README file.
Below are the ensemble results, which roughly resemble the picture in the README but with a noticeably larger uncertainty region.
I was wondering if the training/evaluation parameters were perhaps incorrect and led to this discrepancy. The code for reproducing these results is available on my fork, I made minimal changes just to run the program since there were some errors initially.