learning-dl

learning-dl copied to clipboard

warpctc error: AttributeError: 'module' object has no attribute 'WarpCTC'

1

hello , I tried to run you code about LSTM-CTC, but warpctc doesnt work well .I wonder if you met this problem when run the mxnet with warpctc,I'm looking forward...

qieaaa

can u provide more information about how to prepare data for triplet loss?

4

Any one know how to prepare data for triplet loss?

akturtle

Hello , Thanks for your share of code. However,on my training with devs = [mx.gpu()],it seems that my gpu load keeps a low value below 25%. And since I am...

Arielnabai

i write a huffman softmax in mxnet, but i found it was not efficient in training process. Because in mxnet network graph is changeless，that is we should update all network...

superclocks

when i run infer_cnn_ocr.py after trained ,all of the result is 0. gen captcha: 8183 [ 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1] [ 0.1 0.1 0.1...

abner2015

只需要将多个网络参数，i.e. conv1_weight, 给相同名字就可以了么？先谢谢～

lilhope

Hi. I wanna know why there is no "label" input in "siamese network" ? And how you input the label data?

MoyanZitto

I implement end-to-end chinese plate recognition base on your code

4

https://github.com/szad670401/end-to-end-for-chinese-plate-recognition

szad670401