Mismatching model on the MS2Deepscore V2.0 Zenodo repository?

[AharoniLab]

Hi,

Thanks for inspiring work.

Following the basic instructions for using a pretrained model to compute spectral similarities (https://github.com/matchms/ms2deepscore/tree/main) and loading the 'ms2deepscore_model.pt' model from here: https://zenodo.org/records/10814307

...I get the following warning:

The model version (0.5.0) does not match the version of MS2Deepscore (2.0.0), consider downloading a new model or changing the MS2Deepscore version

Could you kindly clarify?

Also, a slightly less generic file name might be helpful

Thanks,-)

[niekdejonge]

Hi @AharoniLab,
This was a mistake on our end, thanks for notifying us! We forgot updating the version of MS2Deepscore before training that model, which results in this warning.
However, this model should actually work with version 2.0.0 (and not 0.5.0)
So, you can safely use the model and we will fix the warning for future releases of the model.

[AharoniLab]

Hi Niek,
Thanks a lot for updating. The model indeed seems to be able to handle mixed-mode data, to a certain extent (more on that in another thread). I do however, get an error message when trying to setup Monte Carlo predictions (i.e. inference) when using the same mentioned model:

MM_similarity_measure = MS2DeepScoreMonteCarlo(model, n_ensembles=10)
_Monte Carlo Dropout is not supposed to be used with a model where dropout-rate=0_

That model has indeed a dropout rate of zero, but was it even designed to be used with this certainty estimation approach?

Thanks, Nir.