I tried walking into this with my Walk with fastai2 notebooks (covering the main differences in the early notebooks) 
Any good suggestions for tools to label images for segmentation?
1 Like
I will read that, thanks
How do we use our own dataset?
1 Like
My best experience with this has been the mid-level API, especially if your directory structure and sample files are different than the high-level use cases!
why is it called fit_one_cycle vs fit ? what is a cycle ?
1 Like
is very helpful
2 Likes
Look above the discussion started by @maya 
We will cover that next week, but by using the docs, try to figure out how the methods grabbing the data work
1 Like
Take care people! Make sure your actions “flatten the curve”.
2 Likes
https://forums.fast.ai/t/lesson-1-official-topic/65626/233 and replies to it
We will learn that in a few lessons, it’s a specific kind of way of fitting, that’s all you need to know at first.
This was a great first lesson 1! Thank you Fast.ai team! So excited to continue learning together. Stay healthy all!
See you all next week! Loved this online lesson because I got to cuddle with my dog the whole time 
For those of you wanting to run the graphviz code in the notebooks (and you are doing so on your own linux box) … do this:
sudo apt-get install graphviz
2 Likes
fit is more for the same hyper parameters (settings for training) over time. fit_one_cycle changes these over time.
Thank you very much Rachel & Jeremy and Sylvain and others
The “one cycle” policy is something that Sylvain, Leslie, Jeremy, and other Fastai folks empirically determined to be a way to rapidly train a deep learning model. See Sylvain’s blog post: https://sgugger.github.io/the-1cycle-policy.html
4 Likes
THANK YOU SO MUCH , great first lesson
2 Likes