seq2seq_chatbot_new icon indicating copy to clipboard operation
seq2seq_chatbot_new copied to clipboard

Published 20 hours ago • verified publisher icon lc222

InvalidArgumentError (see above for traceback): Found Inf or NaN global norm. : Tensor had NaN values

Open aguang1201 opened this issue 6 years ago • 0 comments

每次训练到1000step的时候就报这个错误. 出错是model.py中,下面代码的最后一句: optimizer = tf.train.AdamOptimizer(self.learing_rate) trainable_params = tf.trainable_variables() gradients = tf.gradients(self.loss, trainable_params) clip_gradients, _ = tf.clip_by_global_norm(gradients, self.max_gradient_norm) 貌似是梯度消失吧.把学习率从0.0001改成0.001还是报错. 求指导

aguang1201 avatar Oct 20 '18 02:10 aguang1201