Zhaolong Xing

I've been busy lately. I'll give you a detailed description later

@KangolHsu It's recommended to look at the new added layers.

@KangolHsu if you want know the details of the auto pruning , refer [here](https://github.com/PaddlePaddle/Mobile/tree/develop/model_compression/pruning)

@KangolHsu This pruning cuts the paramters a in a random way, though it can be accelerated by sparse matrix multiplication, but , it's hard to speed up on embedded devices....

Can you post a detailed log?

The result model is sparse, so the gzip can compress the model to a smaller model.

非常抱歉，此repo不再维护。

Thank you very much for your suggestion. It would be better if you could submit a pr to fix this.

I made a change to merge_model script. For we need model config(`.py`) instead of model `prototxt`, thus a tool likes `converter --model_config mobilenet.py --model_parameters mobilenet.tar.gz --with_rounding True --merge_batch_normalization True` is...