coldgemini

training convergence issue

7

when I train from scratch except for the VGG model part, the loss is always around 1500 and not going any lower after one night training in one GPU. Is...

does your tool support twrp recovery roms or does it only support the official fastboot rom? thanks.

ndc coord -1 to 1 in cascaded shadowmap

2

glm::vec3 frustumCorners[8] = { glm::vec3(-1.0f, 1.0f, -1.0f), glm::vec3( 1.0f, 1.0f, -1.0f), glm::vec3( 1.0f, -1.0f, -1.0f), glm::vec3(-1.0f, -1.0f, -1.0f), glm::vec3(-1.0f, 1.0f, 1.0f), glm::vec3( 1.0f, 1.0f, 1.0f), glm::vec3( 1.0f, -1.0f, 1.0f), glm::vec3(-1.0f,...

support framebuffer_fetch

1

does glslang currently support any of these framebuffer_fetch extensions? is there a way to parse built-ins like gl_LastFragData by glslang? thanks.  

build option for GPU version

2

hi, Do you have the build options for building SimpleElastix with GPU capability? the default is cpu only. Thanks

value cannot be converted to type double without overflow:

1

Is it required to install specific pytorch or do some trick to get rid of this? ==> Preparing data.. Epoch: 0 loc_loss: 0.116 | cls_loss: 3791.763 | train_loss: 3791.878 |...

where to get the initial net.pth?

1

where to get the initial net.pth? from model zoo?

high learning rate in later stages

1

Hi, why did you use much higher (4X) learning rate from 2 stages and leave only the first stage a small learning rate? If it's kind of a fine tune...

so is this actually code for the 2018 paper or 2015 paper?

test bounding cube

1

is there a way to test visibility of a bounding cube instead of testing all the composing triangles of the bounding cube?? Thanks