jph00

62 ms·

781.
Introducing Pytorch for fast.ai(fast.ai)
10 points by jph00 9y ago|0 comments
782.
by jph00 9y ago
Yes but at this stage they're really toy examples. IIRC the best sklearn result they showed on this benchmark had 20% error, whereas the little WRN network was 4%.
783.
by jph00 9y ago
I don't understand why this seems to be getting so much attention. There are plenty of small image datasets around, and wide recognition of the issues with MNIST. I see no evidence at all that this particular dataset is better than MNI
784.
Notes on state of the art techniques for language modeling(fast.ai)
3 points by jph00 9y ago|0 comments
785.
Notes on state of the art techniques for language modeling(fast.ai)
1 points by jph00 9y ago|0 comments
786.
Diversity Crisis in AI, 2017 edition(fast.ai)
1 points by jph00 9y ago|0 comments
787.
by jph00 9y ago
As others have mentioned, installing anaconda is fine. The question volume improves after lesson 3 when we get an audience mic.
788.
by jph00 9y ago
Yes it'll be online a couple of months after completion, or you can apply to join remotely as an international fellow.
789.
by jph00 9y ago
Also the python distribution in question (anaconda) is becoming standard for data science - pytorch, for instance, (covered in part 2) assumes you're using anaconda for its default install method.
790.
by jph00 9y ago
Yes, you should absolutely be able to tackle those tasks after doing the course.
791.
by jph00 9y ago
The very 1st link of the article is a link to part 1. I'll also link the text 'part 1' in that paragraph to make it more clear.
792.
by jph00 9y ago
I had high hopes for them, but it turns out it just puts a tiny 'i' icon in the corner of the screen, with no way (AFAICT) to draw attention to it. When we redo part 1 in October, we'll try to find a way to incorporate a clea
793.
by jph00 9y ago
If you stick to the setup approach in the videos (using the AMIs we created for the course) all the software will be the correct version to go with the notebooks and videos. If you try to use different versions of the software, it'll b
794.
by jph00 9y ago
I somehow forgot to mention in the post - we're teaching a totally updated part 1 course (keras 2, python 3.6, Tensorflow 1.3, recent deep learning research results) starting end of October in San Francisco. Detail here: https:/&
795.
by jph00 9y ago
Yup they'll be re-recorded. Many updates to cover, not just keras 2!
796.
by jph00 9y ago
Apparently they're getting rid of annotations.
797.
by jph00 9y ago
We (including the rest of the learning community) would love to help you - could you post on the forums the problems you are having and we'll try to help sort them out? If you've already posted, please at-mention me there so I don
798.
by jph00 9y ago
You can PM me through the forums to get a wiki login.
799.
by jph00 9y ago
That's a very flattering comparison! :) Such a great book - here's a link for those interested in bayesian methods: http://nbviewer.jupyter.org/github/CamDavidsonPilon/Probabil... In fact that book inspi
800.
by jph00 9y ago
Yup we're starting it in October! :) (Although if folks follow the course and use the AMI or setup scripts provided, all the correct versions will be installed automatically still today.)
801.
UnicodeMath(blogs.msdn.microsoft.com)
3 points by jph00 9y ago|0 comments
802.
by jph00 9y ago
There's a docker image available : https://github.com/deeprig/fastai-course-1 . This is from the same excellent person who created http://crestle.com , which has all the notebooks and data files for th
803.
by jph00 9y ago
Oh and about the teaching philosophy video - until we posted that we had quite a few students express confusion about the top-down approach. After posting it, we've received a lot of positive feedback about it. I understand it's n
804.
by jph00 9y ago
The move from platform.ai to files.fast.ai could have been communicated better - sorry this impacted you. (We tried to highlight everywhere we could, but we can't change the video itself on YouTube unfortunately.) We're redoing th
805.
Deep Learning for Coders – Launching Deep Learning Part 2(fast.ai)
716 points by jph00 9y ago|91 comments
806.
by jph00 9y ago
I love J. It's elegant, expressive, and concise. Just like regex, it looks like line noise at first, but that is because it's designed to get its job done as efficiently as possible. I think it would be a wonderful way to write de
807.
by jph00 9y ago
Well OK... I don't know why you think those are hard (let alone cutting edge - PSRL isn't new), but it seems important for you to feel like you understand things other people aren't smart enough to, so there's probably n
808.
by jph00 9y ago
> For what it's worth, Musk also spent a lot of time discussing nearer-term implications like job displacement. That's great to hear - I honestly had no idea. Which suggests that the media doesn't have as much interest i
809.
by jph00 9y ago
There's nothing hard to understand about RL. I'm not sure where you get that idea - if you find it hard, perhaps you just need to look at some different way. Karpathy summarizes the differences with regular supervised learning in
810.
by jph00 9y ago
Yes exactly. We really hope to see this effort made. It's a fixable problem, but often in history great inequality is only fixed by violent revolution, which I don't want to see happen in my (or my daughter's) time!

More ›