Are there anyone from Ukraine? I’m from Lviv and would like to meet with someone(online or offline) for discussions and etc. If you are interested - just post something in this thread:)
Hi, I’m from Kiev
Kiev +1. Hello from UCU Summer School!)
Hi again, was a little busy last week.
It looks like we have at least a couple of people here interested in study group.
Let’s agree on format and time.
I was planning to propose to have a regular meetup on Sunday evening(like 9 or 10 p.m.) using hangout, how does it sound you all?
Do you want to make conference call on sundays?
Yeah, why not? if you take into account that lectures are on different days during workweek weekend is our only option for a stable schedule.
But I’m open for proposal if you have better arguments and/or a lot of other people also want to have weekend free of studying.
No, i just dont understand for what should we do this. If you have a problem or a question it is much better to write ot in the chat or open the thread, because all the information will be available for each other.
Sunday 9 p.m. works for me.
What benefits I can see:
- Repeat makes perfect - iterative repeating is proved to have really big impact on quality of learning
- Usually, there are at least a couple of things which people miss during lesson and this type of discussion helps to fix this
- Eye to eye communication
- Discussion about some in-depth things - related papers, implementation details, etc
If these reasons aren’t good enough for you and you don’t see any other benefits - maybe study group format just isn’t for you.
P.S. One of the example what study group meetups are you can see in TWiML study group videos https://www.youtube.com/watch?v=kKjgL4vJ8ns&list=PLILZm3MRkvH8Tfx91Z0CHtYluSOTEJdkr&index=1
Ye that sounds great! If you want to host meetup where you will be disassembling the material one more time, then it can work
Hey! How is your learning going on?
Let’s agree on time/format.
9.pm Sunday(can probably move to 10 p.m. ) via somethink like zoom or hangouts(I’ll check what’ easier to use in free mode).
I’ll try to prepare some note about first lesson so we can have something to start with, but since first lesson had a lot of introductions and etc, I expect discussion to go off topic sometimes.
If you have any particular notes or question you’d like to discuss - feel free to post them here.
Also, please inform me if you’ll be joining us on Sunday, so I’ll know that this isn’t just one of my whims that nobody need
Could you be so kind as to add me also to the chat. I would like to participate in Sunday discussion with respect to the first lesson.
I’ll create new chat for everyone interested, so we won’t clutter it with discussion during live course.
Reminder! We’ll have an online meetup in 2 hours(at 9:00). I’ll create private chat for the meetup at 8:30 or so.
Currently we have 5 participants: @dfilippov @leac @dhira108 @vitalik @liberus
If I missed someone or you won’t be able to attand - just write it here or PM me!
@liberus Please add me as well
Please add me as well
Unfortunately, we just finished, but I’ll add you next time!
Ok, just a quick links summary we shared during meeting:
- https://arxiv.org/abs/1810.10180 - new paper about using feed-forward network as an optimizer, maybe we’ll see it’s implementation inside fast.ai till the course end? @jeremy - FYI, may be interesting for you.
- About one cycle policy https://sgugger.github.io/the-1cycle-policy.html , also might be worth checking his other articles as well
- Some interesting stuff about how good machines are at image recognition http://karpathy.github.io/2014/09/02/what-i-learned-from-competing-against-a-convnet-on-imagenet/
spoiler: Karpathy got an error rate of 5.1% on small subset of imagenet, he outperformed trained model he compared with by 1.7%, which IMO is small enough, that we can say machines are already on human level at recognition tasks
- How to download images from google(notebook by Jeremy): https://github.com/fastai/course-v3/blob/master/nbs/dl1/download_images.ipynb
same but from another author: https://www.pyimagesearch.com/2017/12/04/how-to-create-a-deep-learning-dataset-using-google-images/