Closed GraphicsHunter closed 5 years ago
train a deeper NN need much ops, which require a lot of time. so , if you choose Kann , emm...some kind of inefficiency
It doesn't use GPU. For large matrix multiplication and 2d CNN, GPU is much faster than any possible CPU implementations.
Hi, I was wondering about the statement in the README - "KANN is not as scalable, but it is close in flexibility, has a much smaller code base and only depends on the standard C library.".
Why isn't KANN scalable and why isn't it it suitable for training deeper networks?