0002_fastai_saving_a_basic_fastai_model

This is a minimal example showing how to train a fastai model on Kaggle, and save it so you can use it in your app.

# Make sure we've got the latest version of fastai:
# !pip install -Uqq fastai

what to import to handle vision problems in fastai

First, import all the stuff we need from fastai:

from fastai.vision.all import *

how to download and decompress datasets prepared by fastai

This is a dataset of cats and dogs

path = untar_data(URLs.PETS)/'images'

how to tell it is a cat by reading filename

We need a way to label our images as dogs or cats. In this dataset, pictures of cats are given a filename that starts with a capital letter:

def is_cat(x): return x[0].isupper()

how to create dataloaders with from_name_func

Now we can create our DataLoaders:

dls = ImageDataLoaders.from_name_func('.',
    get_image_files(path), valid_pct=0.2, seed=42,
    label_func=is_cat,
    item_tfms=Resize(192))

how to create a pretrained model with resnet18 and error_rate; how to fine tune it 3 epochs

… and train our model, a resnet18 (to keep it small and fast):

learn = vision_learner(dls, resnet18, metrics=error_rate)
learn.fine_tune(3)

/Users/Natsume/mambaforge/lib/python3.9/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead.
  warnings.warn(
/Users/Natsume/mambaforge/lib/python3.9/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

0.00% [0/1 00:00<?]

epoch	train_loss	valid_loss	error_rate	time

20.65% [19/92 00:12<00:48 0.7980]

how to export model to a pickle file and download it from Kaggle

Now we can export our trained Learner. This contains all the information needed to run the model:

learn.export('model.pkl')

Finally, open the Kaggle sidebar on the right if it’s not already, and find the section marked “Output”. Open the /kaggle/working folder, and you’ll see model.pkl. Click on it, then click on the menu on the right that appears, and choose “Download”. After a few seconds, your model will be downloaded to your computer, where you can then create your app that uses the model.

how to convert ipynb to md

from fastdebug.utils import *
import fastdebug.utils as fu

ipy2md()

[jupytext] Reading /Users/Natsume/Documents/fastdebug/nbs/2022part1/0002_fastai_Saving_Model_fastai.ipynb in format ipynb
[jupytext] Writing /Users/Natsume/Documents/fastdebug/nbs/2022part1/0002_fastai_Saving_Model_fastai.md
cp to : /Users/Natsume/Documents/divefastai/Debuggable/jupytext
move to : /Users/Natsume/Documents/fastdebug/mds/2022part1/

[NbConvertApp] Converting notebook /Users/Natsume/Documents/fastdebug/nbs/2022part1/0002_fastai_Saving_Model_fastai.ipynb to markdown
[NbConvertApp] Writing 4849 bytes to /Users/Natsume/Documents/fastdebug/nbs/2022part1/0002_fastai_Saving_Model_fastai.md

move to : /Users/Natsume/Documents/fastdebug/mds_output