oking9

Results 3 comments of oking9

确实, 老师这里没有减. 刚好我也看到这里了. 我觉得没有减的话放缩范围就变成了[min/scalar,1+min/scalar], 原LSTM输出隐藏状态h受tanh函数影响取值范围为[-1,1], 但是例子中最后有线性层将LSTM输出放大, 所以即使没有减也不太会影响性能.

Hi @unixpickle . I find that MiniImageNet source url in fetch_script is already invalid and I tried to find another source on ImageNet website, and I do find one. (http://www.image-net.org/challenges/LSVRC/2012/dd31405981ef5f776aa17412e1f0c112/ILSVRC2012_img_train.tar)....

yes, the multi thread of cpu. I am not an expert in js, if its impossible then sorry, but i would like to assume there is a multi thread option...