Akilesh Badrinaaraayanan

Hello, can you please point to the link of the dataset ?

Hi Matt, Thanks for your suggestion. I started with the "gradient bug" branch, I noticed a small error in resize function (It should be tf.image.resize_images(inputs, [scale_height, scale_width]) and not tf.image.resize_images(inputs,...

Hi, I was running experiments with increased grid size and in some cases the action probability values become so skewed that one particular value is almost one and the rest...

I am using A2C. Cross entropy becomes zero because: say action_prob [p1, p2] (here p1 is order of 10^(-20) that is close to 0, and p2 is close to 1)...

Thanks for the prompt response. Will try the above-mentioned fix. Looking forward to the Python port.