VGG16 in Theano #99
VGG16 in Theano #99
Conversation
… added a test for that, and a comparaison with the googlenet model. Remains a problem with de balanced vgg
|
Eesh, I guess the circle CI stuff is still running on PRs... I will try to fix that and review this ASAP. |
|
I have to say that i don't really see how to fix it ... I saw that there was some circle CI code on your last PR. If there is anything I can do, tell me and i'll try to make it work. |
|
I will fix it and try to rerun the test (I think I can do that) - do you On Mon, Apr 18, 2016 at 4:33 PM, Simon notifications@github.com wrote:
|
|
Ok great ! Yes I think it is. I hope i did everything right ! |
|
Hi Kyle, I saw that you fixed the CircleCI stuff, so I updated my PR so that it passes the tests. Therefore I think its ok, let me know is there are still problems. |
|
OK awesome. I am pretty sure it is good to go but I will give it a once On Wed, Jul 6, 2016 at 8:36 AM, Simon notifications@github.com wrote:
|
|
The only thing I see from this glance over is squashing the commits into On Wed, Jul 6, 2016 at 9:45 AM, Kyle Kastner kastnerkyle@gmail.com wrote:
|
… added a test for that, and a comparaison with the googlenet model. Remains a problem with de balanced vgg fixed a mistake concerning the balanced vgg model finalizing the code to prepare the PR, and added the vgg classfier to the init file added a circle.yml file in order to pass the circleCI tests Update balanced_vgg.py Added circle.yml from scikit-learn, [doc skip] Update circle.yml added the protobuf dependency fixed a typo
|
I can also squash it down (I think) if you want. Just let me know. |
|
OK - I will try and squash this down and merge in the next few days (if you don't get to it @Simon-Prevoteaux ) it looks pretty good to me! |
|
|
||
| def test_googlenet_classifier(): | ||
| """smoke test for googlenet classifier""" | ||
| print 'testing the googlenet classifier' |
There was a problem hiding this comment.
py3 should be print( ) but we probably don't need this
2 participants
Added a classifier class based on the VGG16 model, a test file for the vgg and a comparaison between the vgg model and the googlenet.