Topaz precision curve oscillates over epochs

[giax]

Dear All,

I am trying to use Topoaz to find more views of my particle and to improve the resolution.

Currently, I am training using a clean subset of ~30000 “good” particles and 6000 micrographs.

The precision curve oscillates over the epochs while the AuPrc seems to reach a plateau.

How should I interpret these results ?

Is this going to work ?

With many thanks,

GIA

[CryoEM2]

as long as average precision is on the rise and flatlines, the job has worked and should yield expected results. I typically train with 5000 or less particles, and train individual models for different views and combine the results of several particle picking exercises then remove duplicates.