I wonder if there are any experimental results for this implementation.
lynlynlyn opened this issue · 1 comments
lynlynlyn commented
Thank you very much for the released code.
I wonder if there are any experimental results for this implementation.
g1910 commented
I could achieve an accuracy close to 91% with this implementation. I didn't spend a lot of time to optimize the training. If I remember correctly, the paper did not provide all the fine details to implement it identically. Please feel free to fork this and modify it to see if you can achieve better results.