Skip to live by clicking on “Live” on the bottom of the video feed
It’s live right now. Just click on the 'Live" icon below the video feed
This is a wrong link…the live stream is here - https://www.youtube.com/channel/UCxdAk_H3vzci9ojgtFKjO2g/live
This is where we type our questions?
Here is the thread for questions regarding today’s workshop - In-class discussion: Introductory workshop
Hi Jeremy,
I have been unable to see or hear live stream, could be bad internet. I wish I will get to see video later and also get instructions for set up in wiki somewhere. Looking forward to learn and share.
If all videos will be posted within 24 hours, that’s going to be perfect for us Europeans. I’m actually finishing today’s workshop just now, I simply paused it and I’m now able to watch it until the end. Perfect!
2 Likes
Will we be using AWS instances or Paperspace throughout this course session or both
?
The video is default uploaded to youtube. After the live stream ends it just gets auto uploaded. I just checked ill provide the link here: https://www.youtube.com/watch?v=kjjw4VkVgy8
2 Likes
I edit it and upload an optimized version the day after a lesson. It’s uploading now and I’ll add it to the wiki thread when it’s done.
3 Likes
In order to get the most out of the course jeremy suggests to watch them live as in class participation is a very major point of this international fellowship program 
2 Likes
Oh ok…
1 Like
Hi! Recently researchers from Google published a paper about new activation function they call Swish.
The formula is simply sigmoid(x) * x. The paper shows that it matches or outperforms ReLU activation function in nearly every experiment. Paper.
Read more about it here
5 Likes
I found that paper interesting too. Apparently there is prior art (albeit with the less exciting name of SiLU).
See here: https://arxiv.org/pdf/1702.03118.pdf
2 Likes
There is some debate about Swish performance, as you can see here: https://www.reddit.com/r/MachineLearning/comments/77gcrv/d_swish_is_not_performing_very_well/
1 Like
@jeremy are we to apply the skills, that we learn on a weekly basis in the fast.ai course, to the projects that we applied with and deliver a completed project by the end of the course?
Ideally, that would be an excellent goal! In practice, many people will find that they’ll have plenty more to do once the course if finished - and the forum will keep running so that people can help each other on these projects.
4 Likes
thanks