Hi!
I have been trying the pr2_tensorflow_example, but it appears to have very poor convergence. I am currently fiddling with the configurations in order to try to get better results. I have not tried yet the caffe example, so I wanted to ask if it is behaving in the same way, or is there something wrong?
(I am attaching a picture of the experiment after 10 iterations, the cost appears to remain flat the whole time, except between the first and second iterations, which I found to be a bit odd)
Many thanks for your attention in advance!

best,