Hi,
It seems like there is support for the use of the included general (pretrained) models for Topaz denoise - is this right? If so, would it be possible to add support for use of the general model for picking? (these are available in v0.2.3 of Topaz)
Cheers
Oli
There is an option for “pre-trained initialization” in the cross validation job, but nor in the normal Train job - maybe that is it?
Hi @olibclarke,
These should be added in the next release of cryoSPARC.
Both training jobs will feature the Use pretrained initialization parameter which initializes the model to a pretrained one before training on top of that.
The extraction job will feature a Select pretrained parameter which will allow the selection of any of the pretrained models provided by Topaz.
Regards,
Jay Yoo
hi @jyoo
I was just checking to see if the pretrained models for particle picking were added into the 2.14.0 version?
thanks
Mike
edit: I rebooted the system and it updated but I’m habing the same issue and @olibclarke below
Hi @stephan,
I just tried this, but although the option for selecting a general model is there now in Topaz Extract, it still seems to require the user to input a model - is this intended?
Cheers
Oli
Hi @olibclarke,
This is a bit of a mishap. A current workaround is that if any model, even one that was minimally trained, is input alongside a pre-trained model selection aside from “None”, the job will use the selected pre-trained model and ignore the input model.
Regards,
Jay Yoo
