Surya Prakash Mishra

The 4th page of the paper states > The pooling parameter pk can be manually set or learned since this operation is differentiable and can be part of the back-propagation....

@Zhaopudark Dilation basically means to leave out features (tokens or pixels) in between two features which will be convolved together. If you keep both dilation_rate and strides greater than 1...

@chunduriv In the [documentation ](https://www.tensorflow.org/api_docs/python/tf/keras/layers/Conv1DTranspose) it clearly mentions under `dilation` > Currently, specifying a dilation_rate value != 1 is incompatible with specifying a stride value != 1. And this should...

@gowthamkpr On CPU it throws `Invalid Argument Error` which is right imo. But on GPU it throws `Unimplemented Error` which is confusing IMO.

@fchollet can you take a look now? TIA

Yeah it is now supported by TF v2.0. You can download the tf2 weight file and follow the instruction from [here](https://github.com/bgshih/aster/tree/master/tf2_utils)

Yeah it works in windows environment. You can clone the repo and then install the required packages using pip and run the code finally.

If you are already inside the aster directory you can remove the aster. from imports. Also before running the code you need to build the protos. After you have built...

Hi @crystina-z Sorry to bother, but is there any update regarding this? Would be of a great help for my current project. Thank you in advance.

Hi @LukeWood @innat I would like to work on this! I was thinking of something like this ``` def attach_attention_module(net, attention_module): if attention_module == 'se_block': # SE_block net = se_block(net)...