search

Tencent / PocketFlow

An Automatic Model Compression (AutoMC) framework for developing smaller and faster AI applications.
https://pocketflow.github.io
Other
2.78k stars 490 forks source link

channel pruning for self defined models #275

Open jimengying opened 5 years ago

jimengying commented 5 years ago

当我用--learner=chn-pruned-rmt \ --cpr_prune_ratio=0.40 压缩自定义模型的时候，发现无论我设置的cpr_prune_ratio为多少，最后压缩完成的模型的参数量都是一样的。

jiaxiang-wu commented 5 years ago

The .ckpt model produced by any channel pruned learners does not remove all-zero channels, therefore, the model size stays the same. These all-zero channels will be removed when exporting to .pb and *.tflite models.

huoguangdiandian commented 5 years ago

您好，我是用dcp方法压缩模型的，发现不管设置剪枝率多少，模型大小不变，而且转为.pb格式后模型大小也不变，不知道为什么，求解释。。。谢谢了