"Not enough points" error about eagle HOT 3 CLOSED

Hi! What command are you running? The check that raises this exception is more of a sanity check than a strict requirement, it's just to ensure we're running what is expected - in this case, for NB2 it is expected to have 15284 values since that's the number of models in the search space. The value is specified in a config file, under dataset.total_points and then is checked here: https://github.com/SamsungLabs/eagle/blob/main/eagle/predictors/dataset.py#L88 coming from here: https://github.com/SamsungLabs/eagle/blob/main/eagle/predictors/__main__.py#L520
However, I'm a bit confused since in the current code all the config files seem to be using total_points: null which means that the check is disabled. Please make sure you're running the latest version of our code and if the error still appears please share your command line and config file that you're using. Thanks!

from eagle.

I downloaded it this week, so it should be the latest version. I basically copied the command from the readme file, only editing it because the config file was misspelled as configs/predictors/desktop_gen_nasbench201.yaml instead of configs/predictors/desktop_gcn_nasbench201.yaml.

I checked, and it did seem to have the "total_points" issue as well. I edited that and things seem to be working now.

from eagle.

Hi @GrayGriffin
total_points is set to 15284 in desktop_gcn_nasbench201.yaml
It is expected to be used with the measurement file desktop-cpu-core-i7-7820x-fp32.pickle which also contains 15284 points.

from eagle.