bunzip `fn`, raising exception if output already exists

Decorator to create a simple CLI from `func` using `plac`

Suggest how to improve the setup to speed things up

Reset the state of the inner optimizer.

Add gradient clipping of `clip` during training.

Mimic the behavior of torch.clone for `ImagePoints` objects.

Mimic the behavior of torch.clone for `Image` objects.

Create a Learner for collaborative filtering on `data`.

Initialize the non-batchnorm layers of `m` with `init_func`.

Helper function that returns column names of cont and cat variables from given df.

Convert modified notebooks in `folder` to html pages in `dest_path`.

Equivalent to `image.flow = func(image.flow, image.size)`.

Put `func` with `args` and `kwargs` in `self.flow_func` for later.

Create DeviceDataLoader from `dataset` with `bs` and `shuffle`: process using `num_workers`.

Create a `DataBunch` from `train_ds`, `valid_ds` and maybe `test_ds` with a batch size of `bs`. Passes `**dl_kwargs` to `DataLoader()`

Create a vocabulary from a set of `tokens`.

Function that transform the `datasets` in a `DataBunch` for classification. Passes `**dl_kwargs` on to `DataLoader()`

Calculate the cos and sin of date/time cycles.

Return this images pixels as a tensor.

Return this image pixels as a `LongTensor`.

Return the points associated to this object.

Create an `DataBunch` from self, `path` will override `self.path`, `kwargs` are passed to `DataBunch.create`.

To throw a clear error message when the data wasn't split.

Replace all Capitalized tokens in `x` by their lower version and add `TK_MAJ` before.

Denormalize `x` with `mean` and `std`.

Dice coefficient metric for binary target. If iou=True, returns iou metric, classic for segmentation problems.

Returns a list of all DeviceDataLoaders. If you need a specific DeviceDataLoader, access via the relevant property (`train_dl`, `valid_dl`, etc) as th

Helper function for `anneal_poly`.

Show `show_doc` info in preview window along with link to full docs.

Inline notebook popup for `show_test`

Download images listed in text file `urls` to path `dest`, at most `max_pics`

1 - `accuracy`

Export the state of the `Learner` in `self.path/file`. `file` can be file-like (file or buffer)

Export the minimal state of `self` for inference in `self.path/file`. `file` can be file-like (file or buffer)

Export the minimal state and save it in `fn` to load an empty version for inference.

Extract the keys in `names` from the `kwargs`.

Computes the f_beta between `preds` and `targets`

Only keep filenames in `include` folder or reject the ones in `exclude`.

Keep random sample of `items` with probability `p` and an optional `seed`.

List of label subdirectories in imagenet-style `folder`.

Fit a model following the 1cycle policy.

List of replacements from html strings in `x`.

Tries to convert to float, returns x if it can't

Access the flow-field grid after applying queued affine transforms.

Access the flow-field grid after applying queued affine and coord transforms.

Computes non-background accuracy, e.g. camvid for multiclass segmentation