Have the teams been re-shuffled? I can’t access the spreadsheet anymore to confirm.
1 Like
Is this still the link to the spreadsheet with the team info? https://docs.google.com/spreadsheets/d/1MjY5tVIZy5VtBaDSqMI3wURKEWJQQEqYmnvIiIwcCdY/edit?usp=sharing
For some reason I don’t have permission to see it.
2 Likes
I think it has got to do with Jupyter notebook extensions for aspects like dropdown and creating sections.
@yinterian @jeremy, It seems that forms isn’t loading for few users. Reported by @PranY
Aah I see now. Actually he is doing that after getting the log probabilities. I guess it’s matter of putting another wrapper around this 
preds = np.argmax(log_preds, axis=1) # from log probabilities to 0 or 1
probs = np.exp(log_preds[:,1]) # pr(dog)
1 Like
same here
you can use markdown: http://datascience.ibm.com/blog/markdown-for-jupyter-notebooks-cheatsheet/
ETA: …and there is an extension for collapsible headings etc. http://jupyter-contrib-nbextensions.readthedocs.io/
It looks like I have lost access to the spreadsheet with the group names. =(
Could you please approve it. thanks!
2 Likes
@karlaf, True. But, everyone are seeing the same notebook right from git repo. How come I do not see the indentation ? is there a setting ?
In the previous version of the course, there was an explanation related to dropout: dropout layer kills some information flowing through the neural network during training, but doesn’t do that on validation.
1 Like
Those are tweaks that you do on your jupyter installed on your system. You can customize jupyter the way you want.
You ask people on the forum if you can join their team.
Can you send me a link. Thanks
With respect to the learning rate, I thought you can only set the learning rate for all the layers not individual layers. What’s the intuition for adjusting specific layers vs all the layers?
You can set them for groups of layers (3 groups).
Hello! Is anyone looking to get another person in their group? I attend all the class but couldn’t be in for the class physically today, so I didn’t get assigned a group. I was previously group 5 on last Monday’s class
Is there a function to view all the layers summary in the model ?
Right now we’re doing classification and it’s computing the activations detecting different lines/edges per layer. How does image segmentation work and how is that different than single image/classification?
1 Like
What does precompute=True do?