LJX

Have you solved it？ I tried to load the pretrained resnet50. ``` restore_saver = tf.train.Saver() restore_saver.restore(sess, './models/resnet_v1_50.ckpt') ``` And I met the same problem.

Hello, have you succeeded? I freeze model D to a pb file , but it works bad.

@Yaque It works in fact. I made some mistakes in my code. But after I fixed them, the .pb file works well. I think this may help you https://github.com/LJXLJXLJX/InsightFace_TF_opencv_dnn/blob/master/my_test.py I...

@Yaque All languages that support opencv(4.0.1) works.

> Solved, if you meet the same problem, please re-install your tensor layer to version 1.7 > > ``` > pip uninstall tensorlayer > pip install tensorlayer==1.7 > ``` It...

> > > >  > > > > 我想知道“教师强制：将目标label作为下一个输入”是在干什么？ > > > > > > > > > 加速模型收敛的过程 > > > > > > 使用教师强制和不用教师强制训练出来的最终模型精度是否会不一样呢？ > > 应该会不一样，本来就是随机梯度下降。因为rnn本来就比较难训练，不采用teach...

解决了吗 遇到了类似的问题

@shaqing 没有，我自己尝试改法的无法收敛。

> 确实是在64位Linux下报错，但在32位Linux下运行正常。 问下解决了吗，我在ubuntu虚拟机也遇到这个问题

@HiKapok Thanks ! Your version seems easier to understand