Understanding gradient calculation

I wish I found this thread earlier. After spending 2 weekends cracking my head around how exactly that works and why the code looks like it does, I’ve written a blog post on it. It has a notebook attached so you can play with intermediate steps. Hope it helps somebody if you still don’t get it after reading explanations from @machinethink.

6 Likes

@mariokostelac I’m in the same boat just now as you were. I looked for blog your post, but it seems to be disconnected from the link. Any chance I could take a look at it?

@daveramseymusic The post can be found at Batched backpropagation: connecting the code and the math | model.predict

Thanks @mariokostelac for taking the time to create this super helpful article.

1 Like