fastai v1

Use the W&B callback with fastai v1 to log model topology, losses, metrics, weights, gradients, and predictions.

This documentation is for fastai v1. If you use the current version of fastai, see the fastai page.

For fastai v1 scripts, W&B provides a callback that automatically logs model topology, losses, metrics, weights, gradients, sample predictions, and the best trained model.

import wandb
from wandb.fastai import WandbCallback

wandb.init()

learn = cnn_learner(data, model, callback_fns=WandbCallback)
learn.fit(epochs)

You can configure what data to log through the callback constructor.

from functools import partial

learn = cnn_learner(
    data, model, callback_fns=partial(WandbCallback, input_type="images")
)

You can also use WandbCallback only when starting training. In this case, you must instantiate it.

learn.fit(epochs, callbacks=WandbCallback(learn))

You can also pass custom parameters at that stage.

learn.fit(epochs, callbacks=WandbCallback(learn, input_type="images"))

Example code

The following examples show how the integration works:

• Classify Simpsons characters: A demo to track and compare fastai models. See the step-by-step guide.
• Semantic segmentation with fastai: Optimize neural networks on self-driving cars.

Options

The WandbCallback() class supports several options:

Keyword argument	Default	Description
learn	N/A	The fast.ai learner to hook.
save_model	True	Save the model if it’s improved at each step. It also loads the best model at the end of training.
mode	auto	min, max, or auto. How to compare the training metric specified in monitor between steps.
monitor	None	Training metric used to measure performance for saving the best model. None defaults to validation loss.
log	gradients	gradients, parameters, all, or None. Losses and metrics are always logged.
input_type	None	images or None. Used to display sample predictions.
validation_data	None	Data used for sample predictions if input_type is set.
predictions	36	Number of predictions to make if input_type is set and validation_data is None.
seed	12345	Initialize random generator for sample predictions if input_type is set and validation_data is None.