Hi everyone, and happy new year! I’ve got some great news for you - all the videos from part 1 v2 are now available. They are available in this forum post: Welcome to Part 1 (v2) . You’ll also see that the two v2 forum categories are now publicly available too.
Please don’t share the new videos on high-traffic sites yet, since I don’t want to have lots of new students come here until we’ve got a full new course web-site in place. But feel free to share with friends / work / etc.
BTW, if you’re in the middle of v1 of the course, I’d strongly encourage you to switch to the new v2. It’s a huge step up from v1 in terms of the quality of models that you’ll learn and the amount you’ll be able to do. If you’ve already completed v1, you’ll find most of v2 is new, so it’s worth doing as well (especially lessons 2-4).
About 600 students completed the course in person or through the international fellows program, so there’s lot of folks on the forums ready to help you in your learning journey…
I’m just starting the Fast.ai MOOC this evening. I learned about the course by listening to Rachel Thomas JupyterCon Presentation here in NYC a few months ago.
In this note, are you discouraging new folks from doing v2 “… until we’ve got a full new course website in place.” I’m not clear what category, I would fall into? If it is OK, I’d like to start with V2…
As I’m completely new to this MOOC. (I have done other MOOCs before.) I guess, that I’m asking is their enough materials for V2 to find a good path into the v2 content.
which seems to be a starting point for version 1. It seems to be a set of videos, notebooks, forums, homework assignments, a wiki…
I gather that v2 has some differences and would need new / different notebooks…
I gather that the final website is not complete… (I’m OK with that I think.)
I’m wondering if there is a prototype of a wayfinder, a start page, that points to a set of resources that are good for v2?
Is Welcome to Part 1 (v2) the best starting point? Or is there a good version 1 starting point that I change a few items to get on a good version 2 path. Again I’ve not reviewed most of the wonderful content here still poking around to find the start point. But would love to do the latest learning.
As someone who has taken both versions I’d say it’s mainly a question of whether you’re more interested in Keras/tensorflow or in Pytorch.
The new version includes a lot of very powerful techniques that have been developed over the past year but they aren’t essential to learning deep learning and either course will provide the fundamentals.
My personal recommendation would be to look at both as they provide the same content in slightly different ways. I found that in order to really understand what was going on (and Jeremy recommends this as well) that it’s helpful to watch the videos multiple times, and one way to do this would be to look at both courses.
Either way there’s a lot of overlap in the material and you aren’t missing out on much by doing V1, especially if you follow up by taking a look at V2 and learning pytorch as you do.
I’m partly done with part 1 v1 and seriously considering building my own deep learning box as suggested in part 2(costs around INR 1,70,000 including GTX 1080 Ti 11GB ). I’ve been through some material that suggests waiting sometime before investing in hardware (http://timdettmers.com/2017/12/21/deep-learning-hardware-limbo/ ). would like your opinion on the same… even if the market changes to accommodate Nervana/AMD, is nvidia/CUDA likely to remain the standard for near future, say 12-18 months? or should one wait for better, affordable alternatives?