andravin/wincnn issues and pull requests

#28 - What are the A, G, B of F(2x2,1x3)?

Issue - State: closed - Opened by foreverlms about 2 years ago - 5 comments

#27 - How can calculate F(2x3,3x3)?

Issue - State: closed - Opened by weiqingchen110 almost 3 years ago - 1 comment

#26 - Fixed #25 for Sympy 1.6.2

Pull Request - State: closed - Opened by andravin about 4 years ago

#25 - Broken on Sympy 1.6.2

Issue - State: closed - Opened by asadalam about 4 years ago - 14 comments

#24 - What are the polynomial points for F(14, 3)?

Issue - State: closed - Opened by JIANGJZ over 4 years ago - 6 comments

#23 - Any particular reason to spread denominators into Matrix A B and G?

Issue - State: closed - Opened by oneengineer over 5 years ago - 1 comment

#22 - What are the polynomial points for F(4,5)???

Issue - State: open - Opened by victorygogogo over 5 years ago - 7 comments

#21 - Q: Differences in result.

Issue - State: closed - Opened by cbalint13 over 5 years ago - 2 comments

#20 - Add python3 support.

Pull Request - State: closed - Opened by cbalint13 over 5 years ago - 1 comment

#19 - Is Max Pooling possible in Winograd domain? If not, why?

Issue - State: closed - Opened by aquibjamal almost 6 years ago - 8 comments

#18 - How to calculate the transform matrixes for F(MxN, 5x3) formal.

Issue - State: closed - Opened by lujnan almost 6 years ago - 8 comments

#17 - Number of tiles per channel (to compute convnet layers with F(2 x 2, 3 x 3))

Issue - State: closed - Opened by adam-dziedzic almost 6 years ago - 5 comments

#16 - 8 bit Winograd Convolution?

Issue - State: open - Opened by manojrohit about 6 years ago - 5 comments

#15 - how to get the results order by row

Issue - State: closed - Opened by victorygogogo about 6 years ago - 1 comment

#14 - fast multiplication of complex numbers (from the paper: Fast Algorithms for Convolutional Neural Networks)

Issue - State: closed - Opened by adam-dziedzic over 6 years ago - 4 comments

#13 - F(2,3) transformation matrices is corrected

Pull Request - State: closed - Opened by fatihayar over 6 years ago - 2 comments

#12 - why use the matrix calculate the convolution the multiplication is less ?

Issue - State: closed - Opened by victorygogogo over 6 years ago - 6 comments

#11 - what is the recommanded polynomial points for (7, 3)

Issue - State: closed - Opened by jzhoulon almost 7 years ago - 4 comments

#10 - Choice of Modulo Polynomials

Issue - State: open - Opened by JoWayne almost 7 years ago - 4 comments

#9 - F(3,3) vs F(2,3) accuracy

Issue - State: closed - Opened by chasingegg about 7 years ago - 1 comment

#8 - What are the polynomial points for F(8,3)???

Issue - State: closed - Opened by wuzhijiexia about 7 years ago - 4 comments

#7 - What happens if stride = 2?

Issue - State: closed - Opened by power0341 over 7 years ago - 1 comment

#6 - for stride = 2?

Issue - State: closed - Opened by jian-ke over 7 years ago - 2 comments

#5 - For F(3,3)????

Issue - State: closed - Opened by wuzhijiexia over 7 years ago - 13 comments

#4 - wincnn.showCookToomFilter((0,1,-1), 2, 3)

Issue - State: closed - Opened by wolf1981 over 7 years ago - 1 comment

#3 - Large difference to regular convolution

Issue - State: closed - Opened by Pfaeff almost 8 years ago - 9 comments

#2 - source code about GPU implementation

Issue - State: closed - Opened by NHZlX about 8 years ago - 2 comments

#1 - question: accuracy/numerical stability test?

Issue - State: closed - Opened by kruus over 8 years ago - 2 comments

Ecosyste.ms: Issues

GitHub / andravin/wincnn issues and pull requests