Running pretrained_model

[Arto Pulk]

Hello,

I would like to run more epochs with pretrained model as clearly the training is not converged (imaged attached). How I run the pretrained_model? Do I just use pretrained model in reconstruct script or still need to have other parameters also specified? For example, just pretrained_model:  spisonet.py  --pretrained_model  xx.pt

--epochs 40 or  spisonet.py  reconstruct --pretrained_model  xx.pt --epochs 40 or  spisonet.py reconstruct emd_8731_half_map_1.mrc emd_8731_half_map_2.mrc -- aniso_file FSC3D.mrc --mask emd_8731_msk_1.mrc --limit_res 3.5 --epochs 40 --alpha 1 --beta 0.5 --output_dir isonet_maps --gpuID 0,1,2,3 --acc_batches 2 --pretrained_model  xx.pt

Regards,

Arto

[YUNTAO LIU]

Hi Arto,

I think you need to specify all other parameters, the model does not remember the maps and parameters. 

--
You received this message because you are subscribed to the Google Groups "spIsoNet" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spisonet+u...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/spisonet/2bad7737-cd3f-4e08-9a2a-c592b2b7d642n%40googlegroups.com.

--

Best Regards,

Yuntao Liu,  Postdoc.

California NanoSystem Institute

University of California Los Angeles