Kedar Tatwawadi

@tomrunia Can you take a look? It would be good to have a feature like this as a part of the standard package.

thanks! Here are some instructions (collected at one place for convenience) to run the AOManalyzer locally: https://gist.github.com/kedartatwawadi/4eaddeafccb83497260666ba071e3b3c

In the current version, I have not included Arithmetic coder, as the compression performance obtained by arithmetic coder is exactly equal to the log-loss value obtained at the end. (See:...

Will work on adding the arithmetic encoding/decoding block support soon

Hello, the loss function should give the compression performance (bits/symbol) which would be achieved if arithmetic coding is used. See the updated README and https://web.stanford.edu/~kedart/files/deepzip.pdf for more info. The default...

input.txt is any input you want to compress (make sure to add --vocab option to specify what characters are used in the input.txt you pick) validation.txt can also be any...

Hi all! Thanks for reporting the bug. I am currently on vacation until Thursday, Jan 18th. I will take a look at the bug after the same.

The code was developed on Tensorflow 0.12.1 and is working fine on this version. The bug seems to be coming in the latest stable version TF1.4 I am working on...

apologies. created a requirements.txt.

Based on my experiments, gzip should be able to compress markovity 10 dataset much better than markovity 50. Can you post the command you used for generate the files (with...