Yang Zhou

Results 9 comments of Yang Zhou

pose Jacobian on translation in get3DLocalTransformJac_

@maria-korosteleva , rotation theta for joint i should only cause translation change to child joints of joint i, but not for joint i, if we look into the formula T(i)...

ShapeJac

Thanks for the reply @maria-korosteleva , but the shape will cause the change to shape blended vertices -> cause change to resting joint locations (joint regressor) -> cause change to...

How to avoid "self.current_input_max = F.max(F.abs(x), axis=(1, 2, 3)).mean().asscalar()" to be a node in graph ?

Thanks. I am using it for my own highly customized implementation, I have figured out the error I encountered. It seems the implementation of quantization aware training may have some...

How to avoid "self.current_input_max = F.max(F.abs(x), axis=(1, 2, 3)).mean().asscalar()" to be a node in graph ?

The way to simulate the quantization conv is kind of approximate, which is different from the TensorFlow paper. I think the big difference between the approximating and tensflow, is the...

How to avoid "self.current_input_max = F.max(F.abs(x), axis=(1, 2, 3)).mean().asscalar()" to be a node in graph ?

I am doing experiments on the customized model, wish this approximation could work. I found for my customized model, the fake_bn will significantly slow down the loss descent, once I...

How to avoid "self.current_input_max = F.max(F.abs(x), axis=(1, 2, 3)).mean().asscalar()" to be a node in graph ?

A followup question from my last question. Assuming fused batchnorm is faster, and assuming the training can lead to the same performance. Why all the frameworks directly add an option...

How to avoid "self.current_input_max = F.max(F.abs(x), axis=(1, 2, 3)).mean().asscalar()" to be a node in graph ?

Yes, I have been careful with the "fake_bn", e.g., by-pass-bn. My practical finding is that it is harmful for training, it indeed slows down the model converge (in terms of...

How to avoid "self.current_input_max = F.max(F.abs(x), axis=(1, 2, 3)).mean().asscalar()" to be a node in graph ?

Yes, I am referring to "fused batchnorm" as the TensorFlow paper for the quantization aware training. But your implementation for fake_bn actually is "fused batchnorm"

How to avoid "self.current_input_max = F.max(F.abs(x), axis=(1, 2, 3)).mean().asscalar()" to be a node in graph ?

I have found some TensorFlow quantization aware training code here "https://github.com/tensorflow/tensorflow/blob/e4262fb2fbf1cb33aaea79ff81754d1e92e99af1/tensorflow/contrib/quantize/python/fold_batch_norms.py#L344" " """Computes batch norm correction params. Before batch normalization is frozen: We use batch statistics for batch norm. correction_scale...