Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

CS 20SI: Tensorflow for Deep Learning Research

web.stanford.edu

Post reply on HN
Print view View original

21–30 of 31 posts

Re: CS 20SI: Tensorflow for Deep Learning Research

#21

Post by king_magic »

jonperl wrote: post #16
king_magic wrote: post #12

Earlier quoted context omitted.

IMO the Udacity course is very poorly authored and taught. I would not recommend it to someone who has a primary goal of learning TensorFlow unless you have a pretty deep understanding of deep learning to begin with.

They have a nanodegree class that is much better. I have been taking it the past two months and I highly recommend it https://www.udacity.com/course/deep-learning-nanodegree-foun...

What areas have they improved in?

I was really disappointed by multiple Udacity courses; not saying they can't pull it off, but I've been burned enough by Udacity that I wouldn't consider paying for any courses from them at this point.

Re: CS 20SI: Tensorflow for Deep Learning Research

#23

Post by ConAntonakos »

wavewash wrote: post #6
MrBlue wrote: post #5

Earlier quoted context omitted.

Came here to ask the same question.

From the FAQ: Will lectures be recorded? Because this is a student initiated course, the lectures won't be recorded.

I hope that changes in the future! Looking for a really good course on TensorFlow. Udacity's Self-Driving Car nanodegree is also very hands-on in terms of learning TensorFlow. You build your own mini version.

Re: CS 20SI: Tensorflow for Deep Learning Research

#24

Post by FabHK »

king_magic wrote: post #21
jonperl wrote: post #16

Earlier quoted context omitted.

They have a nanodegree class that is much better. I have been taking it the past two months and I highly recommend it https://www.udacity.com/course/deep-learning-nanodegree-foun...

What areas have they improved in? I was really disappointed by multiple Udacity courses; not saying they can't pull it off, but I've been burned enough by Udacity that I wouldn't consider paying for any courses from them at this point.

Yeah, I took most of the "AI for self driving cars" course taught by Sebastian Thrun himself, and was torn.

He's no doubt an extraordinarily competent researcher in his field, but he was clearly a beginner to Python (either that or he didn't care), and the Python code (while bringing the concepts across) was so poor as to be distracting.

Re: CS 20SI: Tensorflow for Deep Learning Research

#25

Post by vrm »

jimmies wrote: post #14

Side note, the instructor of that course, Chip Huyen, is nothing short of brilliant. Born in Vietnam, at 18 she decided not to go to college, and went traveling with an empty pocket around the world and wrote books. I heard about her getting accepted in many tip-top schools and chose to go to Stanford 4-5 years ago at 22-23 year old. So seems like she is still an undergrad/master student in Stanford or something righ…

She's honestly one of the most amazing people I've ever met. Go chip!!

Re: CS 20SI: Tensorflow for Deep Learning Research

#27

Post by dnautics »

king_magic wrote: post #12
FabHK wrote: post #9

FWIW, Udacity has had a Deep Learning with Tensorflow course for a while. (Note that I'm somewhat ambivalent about Udacity - a lot of it is copy-and-paste stuff, though it does help to get you started.) https://www.udacity.com/course/deep-learning--ud730

IMO the Udacity course is very poorly authored and taught. I would not recommend it to someone who has a primary goal of learning TensorFlow unless you have a pretty deep understanding of deep learning to begin with.

it's very poorly organized. I'd say you get maximal utility out of it if you're like me and for some things (I don't always do this) have to pay for the privilege of being pressured to do something to completion, and are reasonably confident you can figure things out yourself (good maths background, minimally "modestly capable programmer", maybe have tinkered with ML).

Hopefully future iterations will be better.

Re: CS 20SI: Tensorflow for Deep Learning Research

#28

Post by dnautics »

morgangiraud wrote: post #17
hcrisp wrote: post #15

Earlier quoted context omitted.

That is a very helpful primer. What resources did you use to learn that level of understanding in TensorFlow? Just the codebase? If you have any other articles or references to recommend, please do so. Thanks!

Thanks for this feedback! I've just spent some time learning TF on my own using the official doc and made a lot of projects on my Githubs. If you're interested, here is the complete list of my articles so far: How to handle shapes in TensorFlow: https://blog.metaflow.fr/shapes-and-dynamic-dimensions-in-te... TensorFlow saving/restoring and mixing multiple models: https://medium.com/@morgangiraud/tensorflow-saving-res…

wow... Thanks for these links. I find the offical docs of TF are atrocious and I have a lot of trouble figuring it out.

Re: CS 20SI: Tensorflow for Deep Learning Research

#29

Post by rallycarre »

king_magic wrote: post #12
FabHK wrote: post #9

FWIW, Udacity has had a Deep Learning with Tensorflow course for a while. (Note that I'm somewhat ambivalent about Udacity - a lot of it is copy-and-paste stuff, though it does help to get you started.) https://www.udacity.com/course/deep-learning--ud730

IMO the Udacity course is very poorly authored and taught. I would not recommend it to someone who has a primary goal of learning TensorFlow unless you have a pretty deep understanding of deep learning to begin with.

I have taken a 'certification' and a 'class' with Udacity and I wouldn't recommend it to anyone. The courses are shallow and do not challenge the student and the certifications hold no significance.

I prefer EduX courses as they are more like 'classes' and I feel are slightly more respectable to put on a resume.

Post reply on HN