"We have seen what `Tokenizer` or a `Numericalize` do to a collection of texts, and how they're used inside the data block API, which handles those transforms for us directly using the `TextBlock`. But what if we want to only apply one of those transforms, either to see intermediate results or because we have already tokenized texts. More generally, what can we do when the data block API is not flexible enough to accommodate our particular use case? For this, we need to use fastai's *mid-level API* for processing data. The data block API is built on top of that layer, so it will allow you to do everything the data block API does, and much much more."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Going deeper into fastai's layered API"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The fastai library is built on a *layered API*. At the very top layer, there are *applications* that allow us to train a model in five lines of codes, as we saw in <<chapter_intro>>. In the case of creating `DataLoaders` for a text classifier, for instance, we used the line:"
"The factory method `TextDataLoaders.from_folder` is very convenient when your data is arranged the exact same way as the IMDb dataset, but in practice, that often won't be the case. The data block API offers more flexibility. As we saw in the last chapter, we can get the same result with:"
"But it's sometimes not flexible enough. For debugging purposes for instance, we might need to apply just parts of the transforms that come with this data block. Or, we might want to create `DataLoaders` for some application that isn't directly supported by fastai. In this section, we'll dig into the pieces that are used inside fastai to implement the data block API. By understanding these pieces, you'll be able to leverage the power and flexibility of this mid-tier API."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"> note: The mid-level API in general does not only contain functionality for creating `DataLoaders`. It also has the *callback* system , which allows us to customize the training loop any way we like, and the *general optimizer*. Both will be covered in <<chapter_accel_sgd>>."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Transforms"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"When we studied tokenization and numericalization in the last chapter, we started by grabbing a bunch of texts:"
"...and `Tokenizer.decode` turns this back into a single string (it may not, however, be exactly the same as the original string; this depends on whether the tokenizer is *reversible*, which the default word tokenizer is not at the time we're writing this book):"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'xxbos xxmaj this movie , which i just discovered at'"
]
},
"execution_count": null,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"tok.decode(nums_dec)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"`decode` is used by fastai's `show_batch` and `show_results`, as well as some other inference methods, to convert predictions and mini-batches into a human-understandable representation.\n",
"\n",
"For each of `tok` or `num` above, we created an object, called the setup method (which trains the tokenizer if needed for `tok` and creates the vocab for `num`), applied it to our raw texts (by calling the object as a function), and then finally decoded it back to an understandable representation. These steps are needed for most data preprocessing tasks, so fastai provides a class that encapsulates them. This is the `Transform` class. Both `Tokenize` and `Numericalize` are `Transform`s.\n",
"\n",
"In general, a `Transform` is an object that behaves like a function, has an optional *setup* that will initialize some inner state (like the vocab inside `num` for instance), and has an optional *decode* that will reverse the function (this reversal may not be perfect, as we saw above for `tok`).\n",
"\n",
"A good example of `decode` is found in the `Normalize` transform that we saw in <<chapter_sizing_and_tta>>: to be able to plot the images its `decode` method undoes the normalization (i.e. it multiplies by the std and adds back the mean). On the other hand, data augmentation transforms do not have a `decode` method, since we want to show the effects on images, to make sure the data augmentation is working as we want.\n",
"A special behavior of `Transform`s is that they always get applied over tuples: in general, our data is always a tuple `(input,target)` (sometimes with more than one input or more than one target). When applying a transform on an item like this, such as `Resize`, we don't want to resize the tuple, but resize the input (if applicable) and the target (if applicable). It's the same for the batch transforms that do data augmentation: when the input is an image and the target is a segmentation mask, the transform needs to be applied (the same way) to the input and the target.\n",
"If you want to write a custom transform to apply to your data, the easiest way is to write a function. As you can see in this example, a `Transform` will only be applied to a matching type, if a type is provided (otherwise it will always be applied). In the following code, the `:int` in the function signature means that the `f` only gets applied to ints. That's why `tfm(2.0)` returns `2.0`, but `tfm(2)` returns `3` here:"
"Here `f` is converted to a `Transform` with no `setup` and no `decode` method.\n",
"\n",
"Python has a special syntax for passing a function (like `f`) to another function (or something that behaves like a function, known as a `callable` in Python), which is a *decorator*. A decorator is used by prepending a callable with `@`, and placing it before a function definition (there's lots of good online tutorials for Python decorators, so take a look if this is a new concept for you). The following is identical to the previous code:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"(3, 2.0)"
]
},
"execution_count": null,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"@Transform\n",
"def f(x:int): return x+1\n",
"f(2),f(2.0)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"If you need either `setup` or `decode`, you will need to subclass `Transform`. When writing this subclass, you need to implement the actual function in `encodes`, then (optionally), the setup behavior in `setups` and the decoding behavior in `decodes`:"
"Here `NormalizeMean` will initialize some state during the setup (the mean of all elements passed), then the transformation is to subtract that mean. For decoding purposes, we implement the reverse of that transformation by adding the mean. Here is an example of `NormalizeMean` in action:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"(3.0, 5.0, 2.0)"
]
},
"execution_count": null,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"tfm = NormalizeMean()\n",
"tfm.setup([1,2,3,4,5])\n",
"start = 2\n",
"y = tfm(start)\n",
"z = tfm.decode(y)\n",
"tfm.mean,y,z"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Note that the method called and the method implemented are different, for each of these methods:\n",
"So, for instance, you would never call `setups` directly, but instead would call `setup`. The reason for this is that `setup` does some work before and after calling `setups` for you. To learn more about `Transform`s and how you can use them to have different behavior depending on the type of the input, be sure to check the tutorials in the fastai docs."
"To compose several transforms together, fastai provides `Pipeline`. We define a `Pipeline` by passing it a list of `Transform`s; it will then compose the transforms inside it. When you call a `Pipeline` on an object, it will automatically call the transforms inside, in order:"
"And you can call decode on the result of your encoding, to get back something you can display and analyze:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'xxbos xxmaj well , \" cube \" ( 1997 ) , xxmaj vincenzo \\'s first movie , was one of the most interesti'"
]
},
"execution_count": null,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"tfms.decode(t)[:100]"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The only part that doesn't work the same way as in `Transform` is the setup. To properly setup a `Pipeline` of `Transform`s on some data, you need to use a `TfmdLists`."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## TfmdLists and Datasets: Transformed collections"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Your data is usually a set of raw items (like filenames, or rows in a dataframe) to which you want to apply a succession of transformations. We just saw that the succession of transformations was represented by a `Pipeline` in fastai. The class that groups together this pipeline with your raw items is called `TfmdLists`."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### TfmdLists"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Here is the short way of doing the transformation we saw in the previous section:"
"At initialization, the `TfmdLists` will automatically call the setup method of each transform in order, providing them not with the raw items but the items transformed by all the previous `Transform`s in order. We can get the result of our pipeline on any raw element just by indexing into the `TfmdLists`:"
"And the `TfmdLists` knows how to decode for showing purposing:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'xxbos xxmaj well , \" cube \" ( 1997 ) , xxmaj vincenzo \\'s first movie , was one of the most interesti'"
]
},
"execution_count": null,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"tls.decode(t)[:100]"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"In fact, it even has a `show` method:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"xxbos xxmaj well , \" cube \" ( 1997 ) , xxmaj vincenzo 's first movie , was one of the most interesting and tricky ideas that xxmaj i 've ever seen when talking about movies . xxmaj they had just one scenery , a bunch of actors and a plot . xxmaj so , what made it so special were all the effective direction , great dialogs and a bizarre condition that characters had to deal like rats in a labyrinth . xxmaj his second movie , \" cypher \" ( 2002 ) , was all about its story , but it was n't so good as \" cube \" but here are the characters being tested like rats again . \n",
"\n",
" \" nothing \" is something very interesting and gets xxmaj vincenzo coming back to his ' cube days ' , locking the characters once again in a very different space with no time once more playing with the characters like playing with rats in an experience room . xxmaj but instead of a thriller sci - fi ( even some of the promotional teasers and trailers erroneous seemed like that ) , \" nothing \" is a loose and light comedy that for sure can be called a modern satire about our society and also about the intolerant world we 're living . xxmaj once again xxmaj xxunk amaze us with a great idea into a so small kind of thing . 2 actors and a blinding white scenario , that 's all you got most part of time and you do n't need more than that . xxmaj while \" cube \" is a claustrophobic experience and \" cypher \" confusing , \" nothing \" is completely the opposite but at the same time also desperate . \n",
"\n",
" xxmaj this movie proves once again that a smart idea means much more than just a millionaire budget . xxmaj of course that the movie fails sometimes , but its prime idea means a lot and offsets any flaws . xxmaj there 's nothing more to be said about this movie because everything is a brilliant surprise and a totally different experience that i had in movies since \" cube \" .\n"
]
}
],
"source": [
"tls.show(t)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The `TfmdLists` is named with an \"s\" because it can handle a training and validation set with a splits argument. You just need to pass the indices of which elemets are in the training set, and which are in the validation set:"
"If you have manually written a `Transform` that returns your whole data (input and target) from the raw items you had, then `TfmdLists` is the class you need. You can directly convert it to a `DataLoaders` object with the `dataloaders` method. This is what we will do in our Siamese example further in this chapter.\n",
"\n",
"In general though, you have two (or more) parallel pipelines of transforms: one for processing your raw items into inputs and one to process your raw items into targets. For instance, here, the pipeline we defined only processes the input. If we want to do text classification, we have to process the labels as well. \n",
"\n",
"Here we need to do two things: first take the label name from the parent folder. There is a function `parent_label` for this:"
"Then we need a `Transform` that will grab the unique items and build a vocab with it during setup, then will transform the string labels into integers when called. fastai provides this transform, it's called `Categorize`:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"((#2) ['neg','pos'], TensorCategory(1))"
]
},
"execution_count": null,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"cat = Categorize()\n",
"cat.setup(lbls)\n",
"cat.vocab, cat(lbls[0])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"To do the whole setup automatically on our list of files, we can create a `TfmdLists` as before:"
"But then we end up with two separate objects for our inputs and targets, which is not what we want. This is where `Datasets` comes to the rescue."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Datasets"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"`Datasets` will apply two (or more) pipelines in parallel to the same raw object and build a tuple with the result. Like `TfmdLists`, it will automatically do the setup for us, and when we index into a `Datasets`, it will return us a tuple with the results of each pipeline:"
"It can also decode any processed tuple or show it directly:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"('xxbos xxmaj this movie had horrible lighting and terrible camera movements . xxmaj this movie is a jumpy horror flick with no meaning at all . xxmaj the slashes are totally fake looking . xxmaj it looks like some 17 year - old idiot wrote this movie and a 10 year old kid shot it . xxmaj with the worst acting you can ever find . xxmaj people are tired of knives . xxmaj at least move on to guns or fire . xxmaj it has almost exact lines from \" when a xxmaj stranger xxmaj calls \" . xxmaj with gruesome killings , only crazy people would enjoy this movie . xxmaj it is obvious the writer does n\\'t have kids or even care for them . i mean at show some mercy . xxmaj just to sum it up , this movie is a \" b \" movie and it sucked . xxmaj just for your own sake , do n\\'t even think about wasting your time watching this crappy movie .',\n",
" 'neg')"
]
},
"execution_count": null,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"t = dsets.valid[0]\n",
"dsets.decode(t)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The last step is to convert your `Datasets` object to a `DataLoaders`, which can be done with the `dataloaders` method. Here we need to pass along special arguments to take care of the padding problem (as we saw in the last chapter). This needs to happen just before we batch the elements, so we pass it to `before_batch`: "
"`dataloaders` directly calls `DataLoader` on each subset of our `Datasets`. fastai's `DataLoader` expands the PyTorch class of the same name and is responsible for collating the items from our datasets into batches. It has a lot of points of customization but the most important you should know are:\n",
"\n",
"- `after_item`: applied on each item after grabbing it inside the dataset. This is the equivalent of the `item_tfms` in `DataBlock`.\n",
"- `before_batch`: applied on the list of items before they are collated. This is the ideal place to pad items to the same size.\n",
"- `after_batch`: applied on the batch as a whole after its construction. This is the equivalent of the `batch_tfms` in `DataBlock`."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"As a conclusion, here is the full code necessary to prepare the data for text classification:"
"The two differences with what we had above is the use of `GrandParentSplitter` to split our training and validation data, and the `dl_type` argument. This is to tell `dataloaders` to use the `SortedDL` class of `DataLoader`, and not the usual one. This is the class that will handle the construction of batches by putting samples of roughly the same lengths into batches.\n",
"\n",
"This does the exact same thing as our `DataBlock` from above:"
"Let's practice what we just learned on this mid-level API for data preprocessing on a computer vision example now, with a *Siamese Model* input pipeline."
"A *Siamese model* takes two images and has to determine if they are of the same class or not. For this example, we will use the pets dataset again, and prepare the data for a model that will have to predict if two images of pets are of the same breed or not. We will explain here how to prepare the data for such a model, then we will train that model in <<chapter_arch_details>>.\n",
"If we didn't care about showing our objects at all, we could directly create one transform to completely preprocess that list of files. We will want to look at those images though, so we need to create a custom type. When you call the `show` method on a `TfmdLists` or a `Datasets` object, it will decode items until it reaches a type that contains a `show` method and use it to show the object. That `show` method gets passed a `ctx`, which could be a matplotlib axes for images, or the row of a dataframe for texts.\n",
"Here we create a `SiameseImage` object that subclasses `Tuple` and is intended to contain three things: two images, and a boolean that's `True` if they are the same breed. We also implement the special `show` method, such that it concatenates the two images, with a black line in the middle. Don't worry too much about the part that is in the `if` test (which is to show the `SiameseImage` when the images are Pillow images, and not tensors), the important part is in the last three lines."
"The important thing with `Transform`s we saw before is that they dispatch over tuples or their subclasses. That's precisely why we chose to subclass tuple in this instance: this way we can apply any transform that work on images to our `SiameseImage` and it will be applied on each image in the tuple:"
"Here the resize transform is applied to each of the two images, but not the boolean flag. Even if we have a custom type, we can thus benefit from all the data augmentation transforms inside the library.\n",
"We are now ready to build the `Transform` that we will use to get our data ready for a Siamese model. First, we will need a function to determine the class of all our images:"
"Then here is our main transform. For each image, il will, with a probability of 0.5, draw an image from the same class and return a `SiameseImage` with a true label, or draw an image from another class and return a `SiameseImage` with a false label. This is all done in the private `_draw` function. There is one difference between the training and validation set, which is why the transform needs to be initialized with the splits: on the training set, we will make that random pick each time we read an image, whereas on the validation set, we make this random pick once and for all at initialization. This way, we get more varied samples during training, but always the same validation set."
"In the middle level API for data collection, we have two objects that can help us apply transforms on a set of items, `TfmdLists` and `Datasets`. If you remember what we have just seen, one applies a `Pipeline` of transforms and the other applies several `Pipeline` of transforms in parallel, to build tuples. Here, our main transform already builds the tuples, so we use `TfmdLists`:"
"And we can finally get our data in `DataLoaders` by calling the `dataloaders` method. One thing to be careful here is that this method does not take `item_tfms` and `batch_tfms` like a `DataBlock`. The fastai `DataLoader` has several hooks that are named after events: here what we apply on the items after they are grabbed is called `after_item`, and what we apply on the batch once it's built is called `after_batch`."
"Note that we need to pass more transforms than usual: that's because the data block API usually adds them automatically:\n",
"\n",
"- `ToTensor` is the one that converts images to tensors (again, it's applied on every part of the tuple)\n",
"- `IntToFloatTensor` convert the tensor of images that have integers from 0 to 255 to a tensor of floats, and divides by 255 to make the values between 0 and 1."
"And we can now train a model using those `DataLoaders`. It needs a bit more customization than the usual model provided by `cnn_learner` since it has to take two images instead of one. We will see how to create such a model and train it in <<chapter_arch_dtails>>."
"Congratulations — you've completed all of the chapters in this book which cover the key practical parts of training and using deep learning! You know how to use all of fastai's built in applications, and how to customise them using the data blocks API and loss functions. You even know how to create a neural network from scratch, and train it! (And hopefully you now know some of the questions to ask to help make sure your creations help improve society too.)\n",
"\n",
"The knowledge you already have is enough to create full working prototypes of many types of neural network application. More importantly, it will help you understand the capabilities and limitations of deep learning models, and how to design a system which best handles these capabilities and limitations.\n",
"\n",
"In the rest of this book we will be pulling apart these applications, piece by piece, to understand all of the foundations they are built on. This is important knowledge for a deep learning practitioner, because it is the knowledge which allows you to inspect and debug models that you build, and to create new applications which are customised for your particular projects."
