Add Trainer to quicktour #18723
Add Trainer to quicktour #18723
Conversation
|
The documentation is not available anymore as the PR was closed or merged. |
docs/source/en/quicktour.mdx
Outdated
| [`pipeline`] is the easiest way to use a pretrained model for a given task. | ||
|
|
||
| <Youtube id="tiZFewofSLM"/> | ||
|
|
||
| The [`pipeline`] supports many common tasks out-of-the-box: | ||
|
|
||
| **Text**: | ||
| * Sentiment analysis: classify the polarity of a given text. | ||
| * Text generation (in English): generate text from a given input. | ||
| * Name entity recognition (NER): label each word with the entity it represents (person, date, location, etc.). | ||
| * Question answering: extract the answer from the context, given some context and a question. | ||
| * Fill-mask: fill in the blank given a text with masked words. | ||
| * Summarization: generate a summary of a long sequence of text or document. | ||
| * Translation: translate text into another language. | ||
| * Feature extraction: create a tensor representation of the text. | ||
|
|
||
| **Image**: | ||
| * Image classification: classify an image. | ||
| * Image segmentation: classify every pixel in an image. | ||
| * Object detection: detect objects within an image. | ||
|
|
||
| **Audio**: | ||
| * Audio classification: assign a label to a given segment of audio. | ||
| * Automatic speech recognition (ASR): transcribe audio data into text. | ||
|
|
||
| <Tip> |
There was a problem hiding this comment.
I think this introduction in different tasks was quite useful! As a user I feel like I wouldn't necessarily come to the quicktour to learn how to use the library deeply, but rather with a need and a task I'd want to solve. In this case, then showcasing what is supported straight away would be helpful.
There was a problem hiding this comment.
Ok, I think Sylvain also advocated for showcasing what all is supported so I'll keep it. Going to experiment a bit with presenting the tasks in a table :)
docs/source/en/quicktour.mdx
Outdated
| <Tip> | ||
|
|
||
| For tasks that use a sequence-to-sequence model like translation or summarization, use the [`Seq2SeqTrainer`] and [`Seq2SeqTrainingArguments`] classes instead. | ||
|
|
||
| </Tip> |
There was a problem hiding this comment.
Here I'd also show code samples
There was a problem hiding this comment.
Since the code sample is so similar to the Seq2Seq classes, what do you think about just clarifying the tip to say that you can copy the above code and remove Seq2Seq? This way, we can avoid being too repetitive.
| You can customize the training loop behavior by subclassing the methods inside [`Trainer`]. This allows you to customize features such as the loss function, optimizer, and scheduler. Take a look at the [`Trainer`] reference for which methods can be subclassed. | ||
|
|
||
| The other way to customize the training loop is by using [Callbacks](./main_classes/callbacks). You can use callbacks to integrate with other libraries and inspect the training loop to report on progress or stop the training early. Callbacks do not modify anything in the training loop itself. To customize something like the loss function, you need to subclass the [`Trainer`] instead. |
There was a problem hiding this comment.
That's more for the Trainer page, but I'd showcase an example of subclassing each method. I think there's a single one shown for compute_loss, but subclassing requires understanding which inputs/outputs will work and it's not necessarily straightforward for a beginner
This PR makes some edits to the
pipelinesection to focus less on all the tasks (and their definitions) it is capable of. Users probably only need a general representative idea of what it can do, and then they're more interested in diving into how to use the pipeline.I also added a brief section on the
Trainerhere about the basic parameters it accepts and a small explanation of how to customize the training loop behavior to keep the quick tour short. I think theTraineris pretty important to include since a lot of users use it for training, and we also use it in our finetune guides.