Hello all,

Based on the information from lesson 9, it would be beneficial to train models with loss function that uses binary classification. This can be a way to detect ambiguous predictions (eg `1`

that looks like `7`

).

In a dataset with one-hot encoded labels,

- Is it possible to train the model using BCEWithLigitsFlat instead of CrossEntropyFlat?
- How would you convert a dataset from multi-class to multi-label?

For example, using MNIST with `data.c = 10`

,

```
from fastai.vision import *
mnist = untar_data(URLs.MNIST)
data = (ImageList.from_folder(mnist/"training")
.split_none()
.label_from_folder()
.transform(tfms, size=32)
.databunch()
.normalize(imagenet_stats))
learn = cnn_learner(data, models.resnet18, metrics=accuracy)
learn.loss_func=BCEWithLogitsFlat()
learn.fit(1)
```

Returns with `ValueError: Target size (torch.Size([64])) must be the same as input size (torch.Size([640]))`

.

It seems the issue resides in the last FC layerâ€™s `out_features`

.

```
# learn.layer_groups[-1]
Sequential(
(0): AdaptiveAvgPool2d(output_size=1)
(1): AdaptiveMaxPool2d(output_size=1)
(2): Flatten()
(3): BatchNorm1d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(4): Dropout(p=0.25, inplace=False)
(5): Linear(in_features=1024, out_features=512, bias=True)
(6): ReLU(inplace=True)
(7): BatchNorm1d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(8): Dropout(p=0.5, inplace=False)
(9): Linear(in_features=512, out_features=10, bias=True)
)
```

Thanks!