SOURCE CODE: Mid-Level API

Part 1 (2020) Study Groups
202

ok … after playing around with vscode for hours I am finally able to step into the fastai libraries in the debugger :smile:

I had to create a launch.json and add the following config parameter.

“justMyCode”: false

here’s a guide how to create the launch.json

203

Blog #2 is up, it’s a short one, but I think that I’ll have a number of smaller ones that cover just little ideas that wouldn’t work in a bigger blog. This one focuses on the idea of utilizing functions in our DataBlock (such as using a dictionary to transform our y values): The Idea of a Transform

3 Likes
205

hey @muellerzr , just a quick question, actually just a point of clarification,
was going through the Blog #1

Screenshot 2020-03-23 at 12.01.49 PM
Screenshot 2020-03-23 at 12.01.49 PM2192×324 54.8 KB

by that you mean, in case we have ‘n’ #epochs (learn.lr_find(n)), the pre-trained model will be trained on ‘n-1’ epochs on the frozen model and last ‘n’ th epoch will be trained on the unfrozen model.
is that correct?

206

@0tist no, we’ll pass in n epochs for our model to train on while unfrozen, and 1 epoch (the default) to train frozen. We’d also use learn.lr_find() (without n in there) to try to find a good learning rate but in general fine_tune does a great job

207

oh my bad, i meant learn.fine_tune(n)
thanks for the clarification :smiley:

1 Like
208

No problem :slight_smile: There’s two separate params in it for epochs, epochs and frozen_epochs. The latter of which has the default of 1, the prior is what we set when we pass in n epochs.

209

For everyone here, should we have a quick catch up today and discuss what we did this week and do a knowledge share lesson?

Let’s meet at 8pm AEST? Does that work for everyone?

210

If all goes to plan, I should be asleep by then, so perhaps a little earlier if we could? (that’s around 5am my time)

211

yes its a bit early in east cost as well =)

212

Okay how about in 2 hours from now? :slight_smile:

213

I can do that :slight_smile:

1 Like
214

@arora_aman Zoom? Discord? :slight_smile:

1 Like
215

Joining official Zoom now.

1 Like
216

Zoom link please. Can’t locate it

217

We’re on the course zoom link :slight_smile:

218

First blog about the DataBlocks API and a gentle code-first introduction is now available here! :slight_smile:

In this blog I use 5 different examples similar to the “Cats Vs Dogs” introduced in Lesson-1 and use the DataBlocks API to build custom datablocks using different methods.

We also look slightly in to convenience functions such as download_data.

6 Likes
219

can someone summarized what was discussed on zoom call ?=)

220

Sure! :slight_smile:

We just discussed what we worked on in the first week.

@muellerzr mentioned his blog and the work he has been doing around coral detection.

We had @barnacl mention about his work around the DataBlocks API too which involved building a DataBlock for different Kaggle competitions.

I mentioned the DataBlocks API examples which are now available as a blog post.

Then I had to jump to a work call.

3 Likes
221

Hi @arora_aman I read your blog and I’ve learnt some new things, but I noticed some of the examples in which you use get_paths_from_df function, this can be simplified using a ColReader, is there any reason you chose to define the former?

1 Like
222

Thanks! Yes, that simplified example 5. I have added credits to the post :slight_smile:

2 Likes