Need to save training checkpoints after several epoch so next training could start from this one

Hi
i want to train my style..
but i got error as this

tensorflow/core/common_runtime/base_collective_executor.cc:216] BaseCollectiveExecutor::StartAbort Failed precondition: Error while reading resource variable _AnonymousVar95 from Container: localhost. This could mean that the variable was uninitialized. Not found: Resource localhost/_AnonymousVar95/class tensorflow::Var does not exist.
[[{{node loss/block5_pool_loss/dummy_loss/weighted_loss/ExpandDims/ReadVariableOp}}]]
Traceback (most recent call last):

Do you have any idea?

i'm using
Anaconda 3.7
window 10
tensorflow-gpu
i5, gtx1060

i don't know why ...
anyway it is solved by chance...

Hello maajor,
for my master thesis maybe i want to us your plugin for a study. But now I download your code from GitHub and follow your instructions in the documentation.pdf till "click play button". The engine switched automatically in pause-mode and the console was full of the follow errors:

Thread group count is above the maximum allowed limit. Maximum allowed thread group count is 65535. UnityEngine.ComputeShader:Dispatch(Int32, Int32, Int32, Int32) NNPP.ReLU:Run(Object[]) (at Assets/NNPostProcessing/Scripts/NNLayers/ReLU.cs:37) NNPP.NNModel:Predict(RenderTexture) (at Assets/NNPostProcessing/Scripts/NNModel.cs:108) NNPP.NNPostProcessingEffect:OnRenderImage(RenderTexture, RenderTexture) (at Assets/NNPostProcessing/Scripts/NNPostProcessingEffect.cs:32)

I don't know how to fix that. I hope u can help me with this errors.

Best regards Markus

So the good news is that this runs in VR without any modification.

Bad news is that it runs at 2fps on a GTX 1080 ;-)

My hunch is that reprojection would make this tolerable if we could hit 45fps.

So only a 20x speedup!

Supporting single pass rendering would probably buy a 1.5x speedup and might not be too hard. Rendering at half res would probably give a 4x speedup.

I haven't looked at the code. How much scope is there for something completely improbably like a 10x speedup through optimization? I'm guessing that's a fairly tall order.

Hi,
I am trying to follow the steps using fast-style-transfer to train. But when I run
python3 train_fst.py --style chinese_landscape.jp
I got this error:
Traceback (most recent call last): File "train_fst.py", line 6, in <module> from scipy.misc import imsave ImportError: cannot import name 'imsave' from 'scipy.misc' (/Library/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/scipy/misc/__init__.py)
I am using the latest TensorFlow 2.3.0 and recommended scipy 1.4.1. Sounds like imsave is depreciated in newer version of scipy. Do you know how I can workaround this issue?