SGD with momentum in tensor flow

Question

In caffe, the SGD solver has a momentum parameter ( link ). In the tensor stream, I see that "tf.train.GradientDescentOptimizer" does not have an explicit momentum parameter. However, I see that there is an optimizer tf.train.MomentumOptimizer. Is this the equivalent of SGD caffe with momentum optimizer?

Answer 1

Yes it is. tf.train.MomentumOptimizer = SGD + boost