Youtube channel exclusive for fastai

Jeremy did a great job coming up with fastai , his course and the books he made available in github . Since this project got a greater community , why cant some one comeup with explaining deeplearning stuff in laymans term , explaining different projects using fastai . it would have been great . Honestly the documentation part really gets me worrying . Or the least if there was some discord channel or telegram for fastai study group , it would have been great .

you could explain what it is in laymans terms but if you want someone to explain how to use it in laymans terms so simple that your granny would understand them then i think you’re going to be waiting a long time.

i think they’ve done a great job making this as simple as possible and they’ve done a much better job in this regard than any other library i’ve seen so far, but you’re still going to have a lot of learning ahead of you.

you need to be proficient in python if you aren’t already. you need a basic understanding of the maths, a basic understanding of statistics, and a basic understand of what’s going on inside the models so you know where to poke it to make it do the stuff you want. if your knowledge is lacking in any of these areas then you need to go and teach yourself.

then you need to experiment and run examples and make your own projects using the code from the lessons and figure it out when stuff doesn’t work because that’s how you learn the most. when you don’t understand what a function does after jeremy explains it, go and look at the code, or run the example step by step until you do. play with it.

there is a lot to learn. that can’t be helped. that’s life as any kind of programmer or engineer, and you need to get used to teaching yourself.

3 Likes

Taking a print of this qoute for my desk :smiley:

5 Likes

I’m sorry for totally being childish coming up with that question . Honestly , your advice is the only solution for most of the problem . I forgot about kaggle’s contribution in this . Found out most people are solving dl problems using fastai in kaggle . I’ll go through those kernels .

you’ll find loads of useful code to play with on places like kaggle, github, here, everywhere really. us nerds love to share and show off.

just don’t be put off by the amount you have to learn to do this. at some point every single person here knew absolutely nothing about any of it.

if you already know some python or some stats or some maths then you’ve got a headstart, but it all works out the same in the end. we all have to learn it all, step by step.

2 Likes

https://www.youtube.com/channel/UCmKoQOD8uBqsRS8XDdSgrlQ/videos

Check out muellerzr youtube. Maybe this will be close to what you are looking for.

1 Like

As was pointed out, my youtube channel attempts to do this, especially with my Walk with fastai2 series :slight_smile: It’s example based to the point how do we get x data to work. Let me know if that is what you’re looking for :slight_smile: I have a few other topics besides WWF2, such as my fastinference sublibrary and a few other estranged fastai topic videos (like using nbdev)

600 hundred followers only? It seams a bit strange as judging by the video title these are pretty searchable. The problem may be in tags. Have you tried adjusting those? It’s not that hard, the software for it nowadays is free as demand is huge. Here’s one I use https://backlinkvalidator.com/tools/youtube_tags_tools.php. May be helpful

1 Like