Manu Mathew

That flag is only propagated from the iteration quantization_start onwards. Default value of this is 2000. So you will see the effect of quantization from iteration 2000 onwards. You can...

I think NVCaffe doesn't support building on CPU. So caffe-jacinto also may not build on CPU.

label_weights can be used during training in the SoftmaxWithLoss layer. In your SoftmaxWithLoss layer, you can provide an array for label_weights. The number of entries must match the number of...

Hi, I don't think the implementation need to be changed. Feel free to change the code to include which ever layers you would like to include. The DW seperable convolution...

Hi, I am not sure which model. Can you please share the path of the original and the sparse model. Thanks,

There should not be any size different between the original model and the sparse model. In the sparse model, several coefficients are zero - that is all. For example check...

Hi, Your first post said: **I tried to use a snapshot of a sparsified model from caffe-jacinto and to load it in caffe. However, I get an error** Your last...

Okay. Can you try one thing? - train the model in NVIDIA/caffe and check the size. Also see if it can be loaded in BVLC caffe. If not, then you...

Try filing an issue in NVIDIA/caffe. They might be able to help.

Have you seen this? https://github.com/NVIDIA/caffe/issues/436 I also noticed the issue that you filed: https://github.com/NVIDIA/caffe/issues/439 These two seem related.