New model ready for testing!

[github-actions[bot]]

This model is a new incorporation to the Ersilia Model Hub or it has been modified. If you are assigned to this issue, please try it out and ensure everything works! To test a model, first clone it in your local system (ideally, from dockerhub) using the CLI commands:

ersilia -v fetch eosxxxx --from_dockerhub
ersilia serve eosxxxx
ersilia test

The test command will automatically check that the model can handle null outputs and whether it produces consistent results. Please copy here the result of the test command. If it passes, simply close the issue as completed. If it fails, please detail at which step and whether you have taken any steps to solve it. Please tag the original model contributor and one of Ersilia's maintainers for support.

[GemmaTuron]

Hi @miquelduranfrigola input.csv output.csv

The model seems to be working fine, see example input and output files (using docker image of the model). Could you please explain the results? There is huge variability between methods

[miquelduranfrigola]

These are three methods (random forset, gradient boosting classifier, and logistic regression). Scores are not calibrated.

[GemmaTuron]

ok... do we want to maybe calibrate them or st? As it is now it is quite confusing for users if the same molecule scores soo differently between models

[miquelduranfrigola]

This is a good idea. Let's leave this issue open and as soon as we can we will tackle it!