From 16d4acbfdb547cb922361ba07a13de12e1503fb8 Mon Sep 17 00:00:00 2001
From: Steven Liu <59462357+stevhliu@users.noreply.github.com>
Date: Fri, 28 Jan 2022 19:01:37 -0600
Subject: [PATCH] Get started docs (#15098)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* clean commit of changes

* apply review feedback, make edits

* fix backticks, minor formatting

* 🖍 make fixup and minor edits

* 🖍 fix # in header

* 📝 update code sample without from_pt

* 📝 final review
---
 docs/source/index.mdx        |  52 +----
 docs/source/installation.mdx | 211 ++++++++++-------
 docs/source/quicktour.mdx    | 433 +++++++++++++----------------------
 3 files changed, 286 insertions(+), 410 deletions(-)

diff --git a/docs/source/index.mdx b/docs/source/index.mdx
index 3192ce1142..562a94121a 100644
--- a/docs/source/index.mdx
+++ b/docs/source/index.mdx
@@ -12,25 +12,18 @@ specific language governing permissions and limitations under the License.
 
 # 🤗 Transformers
 
-State-of-the-art Machine Learning for Jax, Pytorch and TensorFlow
+State-of-the-art Machine Learning for PyTorch, TensorFlow and JAX.
 
-🤗 Transformers (formerly known as _pytorch-transformers_ and _pytorch-pretrained-bert_) provides thousands of pretrained models to perform tasks on different modalities such as text, vision, and audio.
+🤗 Transformers provides APIs to easily download and train state-of-the-art pretrained models. Using pretrained models can reduce your compute costs, carbon footprint, and save you time from training a model from scratch. The models can be used across different modalities such as:
 
-These models can applied on:
+* 📝 Text: text classification, information extraction, question answering, summarization, translation, and text generation in over 100 languages.
+* 🖼️ Images: image classification, object detection, and segmentation.
+* 🗣️ Audio: speech recognition and audio classification.
+* 🐙 Multimodal: table question answering, optical character recognition, information extraction from scanned documents, video classification, and visual question answering.
 
-* 📝 Text, for tasks like text classification, information extraction, question answering, summarization, translation, text generation, in over 100 languages.
-* 🖼️ Images, for tasks like image classification, object detection, and segmentation.
-* 🗣️ Audio, for tasks like speech recognition and audio classification.
+Our library supports seamless integration between three of the most popular deep learning libraries: [PyTorch](https://pytorch.org/), [TensorFlow](https://www.tensorflow.org/) and [JAX](https://jax.readthedocs.io/en/latest/). Train your model in three lines of code in one framework, and load it for inference with another.
 
-Transformer models can also perform tasks on **several modalities combined**, such as table question answering, optical character recognition, information extraction from scanned documents, video classification, and visual question answering.
-
-🤗 Transformers provides APIs to quickly download and use those pretrained models on a given text, fine-tune them on your own datasets and then share them with the community on our [model hub](https://huggingface.co/models). At the same time, each python module defining an architecture is fully standalone and can be modified to enable quick research experiments.
-
-🤗 Transformers is backed by the three most popular deep learning libraries — [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) and [TensorFlow](https://www.tensorflow.org/) — with a seamless integration between them. It's straightforward to train your models with one before loading them for inference with the other.
-
-This is the documentation of our repository [transformers](https://github.com/huggingface/transformers). You can
-also follow our [online course](https://huggingface.co/course) that teaches how to use this library, as well as the
-other libraries developed by Hugging Face and the Hub.
+Each 🤗 Transformers architecture is defined in a standalone Python module so they can be easily customized for research and experiments.
 
 ## If you are looking for custom support from the Hugging Face team
 
@@ -38,35 +31,6 @@ other libraries developed by Hugging Face and the Hub.
 <img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
 </a><br>
 
-## Features
-
-1. Easy-to-use state-of-the-art models:
-    - High performance on natural language understanding & generation, computer vision, and audio tasks.
-    - Low barrier to entry for educators and practitioners.
-    - Few user-facing abstractions with just three classes to learn.
-    - A unified API for using all our pretrained models.
-
-1. Lower compute costs, smaller carbon footprint:
-    - Researchers can share trained models instead of always retraining.
-    - Practitioners can reduce compute time and production costs.
-    - Dozens of architectures with over 20,000 pretrained models, some in more than 100 languages.
-
-1. Choose the right framework for every part of a model's lifetime:
-    - Train state-of-the-art models in 3 lines of code.
-    - Move a single model between TF2.0/PyTorch/JAX frameworks at will.
-    - Seamlessly pick the right framework for training, evaluation and production.
-
-1. Easily customize a model or an example to your needs:
-    - We provide examples for each architecture to reproduce the results published by its original authors.
-    - Model internals are exposed as consistently as possible.
-    - Model files can be used independently of the library for quick experiments.
-
-[All the model checkpoints](https://huggingface.co/models) are seamlessly integrated from the huggingface.co [model
-hub](https://huggingface.co) where they are uploaded directly by [users](https://huggingface.co/users) and
-[organizations](https://huggingface.co/organizations).
-
-Current number of checkpoints: <img src="https://img.shields.io/endpoint?url=https://huggingface.co/api/shields/models&color=brightgreen">
-
 ## Contents
 
 The documentation is organized in five parts:
diff --git a/docs/source/installation.mdx b/docs/source/installation.mdx
index 0b1a710434..cd48721663 100644
--- a/docs/source/installation.mdx
+++ b/docs/source/installation.mdx
@@ -1,5 +1,5 @@
 <!---
-Copyright 2020 The HuggingFace Team. All rights reserved.
+Copyright 2022 The HuggingFace Team. All rights reserved.
 
 Licensed under the Apache License, Version 2.0 (the "License");
 you may not use this file except in compliance with the License.
@@ -16,183 +16,220 @@ limitations under the License.
 
 # Installation
 
-🤗 Transformers is tested on Python 3.6+, and PyTorch 1.1.0+ or TensorFlow 2.0+.
+Install 🤗 Transformers for whichever deep learning library you're working with, setup your cache, and optionally configure 🤗 Transformers to run offline.
 
-You should install 🤗 Transformers in a [virtual environment](https://docs.python.org/3/library/venv.html). If you're
-unfamiliar with Python virtual environments, check out the [user guide](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/). Create a virtual environment with the version of Python you're going
-to use and activate it.
+🤗 Transformers is tested on Python 3.6+, PyTorch 1.1.0+, TensorFlow 2.0+, and Flax. Follow the installation instructions below for the deep learning library you are using:
 
-Now, if you want to use 🤗 Transformers, you can install it with pip. If you'd like to play with the examples, you
-must install it from source.
+* [PyTorch](https://pytorch.org/get-started/locally/) installation instructions.
+* [TensorFlow 2.0](https://www.tensorflow.org/install/pip) installation instructions.
+* [Flax](https://flax.readthedocs.io/en/latest/) installation instructions.
 
-## Installation with pip
+## Install with pip
 
-First you need to install one of, or both, TensorFlow 2.0 and PyTorch.
-Please refer to [TensorFlow installation page](https://www.tensorflow.org/install/pip#tensorflow-2.0-rc-is-available),
-[PyTorch installation page](https://pytorch.org/get-started/locally/#start-locally) and/or
-[Flax installation page](https://github.com/google/flax#quick-install)
-regarding the specific install command for your platform.
+You should install 🤗 Transformers in a [virtual environment](https://docs.python.org/3/library/venv.html). If you're unfamiliar with Python virtual environments, take a look at this [guide](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/). A virtual environment makes it easier to manage different projects, and avoid compatibility issues between dependencies.
 
-When TensorFlow 2.0 and/or PyTorch has been installed, 🤗 Transformers can be installed using pip as follows:
+Start by creating a virtual environment in your project directory:
+
+```bash
+python -m venv .env
+```
+
+Activate the virtual environment:
+
+```bash
+source .env/bin/activate
+```
+
+Now you're ready to install 🤗 Transformers with the following command:
 
 ```bash
 pip install transformers
 ```
 
-Alternatively, for CPU-support only, you can install 🤗 Transformers and PyTorch in one line with:
+For CPU-support only, you can conveniently install 🤗 Transformers and a deep learning library in one line. For example, install 🤗 Transformers and PyTorch with:
 
 ```bash
 pip install transformers[torch]
 ```
 
-or 🤗 Transformers and TensorFlow 2.0 in one line with:
+🤗 Transformers and TensorFlow 2.0:
 
 ```bash
 pip install transformers[tf-cpu]
 ```
 
-or 🤗 Transformers and Flax in one line with:
+🤗 Transformers and Flax:
 
 ```bash
 pip install transformers[flax]
 ```
 
-To check 🤗 Transformers is properly installed, run the following command:
+Finally, check if 🤗 Transformers has been properly installed by running the following command. It will download a pretrained model:
 
 ```bash
 python -c "from transformers import pipeline; print(pipeline('sentiment-analysis')('we love you'))"
 ```
 
-It should download a pretrained model then print something like
+Then print out the label and score:
 
 ```bash
 [{'label': 'POSITIVE', 'score': 0.9998704791069031}]
 ```
 
-(Note that TensorFlow will print additional stuff before that last statement.)
+## Install from source
 
-## Installing from source
-
-Here is how to quickly install `transformers` from source:
+Install 🤗 Transformers from source with the following command:
 
 ```bash
 pip install git+https://github.com/huggingface/transformers
 ```
 
-Note that this will install not the latest released version, but the bleeding edge `master` version, which you may want to use in case a bug has been fixed since the last official release and a new release hasn't been yet rolled out.
+This command installs the bleeding edge `master` version rather than the latest `stable` version. The `master` version is useful for staying up-to-date with the latest developments. For instance, if a bug has been fixed since the last official release but a new release hasn't been rolled out yet. However, this means the `master` version may not always be stable. We strive to keep the `master` version operational, and most issues are usually resolved within a few hours or a day. If you run into a problem, please open an [Issue](https://github.com/huggingface/transformers/issues) so we can fix it even sooner!
 
-While we strive to keep `master` operational at all times, if you notice some issues, they usually get fixed within a few hours or a day and you're more than welcome to help us detect any problems by opening an [Issue](https://github.com/huggingface/transformers/issues) and this way, things will get fixed even sooner.
-
-Again, you can run:
+Check if 🤗 Transformers has been properly installed by running the following command:
 
 ```bash
-python -c "from transformers import pipeline; print(pipeline('sentiment-analysis')('I hate you'))"
+python -c "from transformers import pipeline; print(pipeline('sentiment-analysis')('I love you'))"
 ```
 
-to check 🤗 Transformers is properly installed.
-
 ## Editable install
 
-If you want to constantly use the bleeding edge `master` version of the source code, or if you want to contribute to the library and need to test the changes in the code you're making, you will need an editable install. This is done by cloning the repository and installing with the following commands:
+You will need an editable install if you'd like to:
 
-``` bash
+* Use the `master` version of the source code.
+* Contribute to 🤗 Transformers and need to test changes in the code.
+
+Clone the repository and install 🤗 Transformers with the following commands:
+
+```bash
 git clone https://github.com/huggingface/transformers.git
 cd transformers
 pip install -e .
 ```
 
-This command performs a magical link between the folder you cloned the repository to and your python library paths, and it'll look inside this folder in addition to the normal library-wide paths. So if normally your python packages get installed into:
-```
-~/anaconda3/envs/main/lib/python3.7/site-packages/
-```
-now this editable install will reside where you clone the folder to, e.g. `~/transformers/` and python will search it too.
+These commands will link the folder you cloned the repository to and your Python library paths. Python will now look inside the folder you cloned to in addition to the normal library paths. For example, if your Python packages are typically installed in `~/anaconda3/envs/main/lib/python3.7/site-packages/`, Python will also search the folder you cloned to: `~/transformers/`.
 
-Do note that you have to keep that `transformers` folder around and not delete it to continue using the  `transformers` library.
+<Tip warning={true}>
 
-Now, let's get to the real benefit of this installation approach. Say, you saw some new feature has been just committed into `master`. If you have already performed all the steps above, to update your transformers to include all the latest commits, all you need to do is to `cd` into that cloned repository folder and update the clone to the latest version:
+You must keep the `transformers` folder if you want to keep using the library.
 
-```
+</Tip>
+
+Now you can easily update your clone to the latest version of 🤗 Transformers with the following command:
+
+```bash
 cd ~/transformers/
 git pull
 ```
 
-There is nothing else to do. Your python environment will find the bleeding edge version of `transformers` on the next run.
+Your Python environment will find the `master` version of 🤗 Transformers on the next run.
 
+## Install with conda
 
-## With conda
+Install from the conda channel `huggingface`:
 
-Since Transformers version v4.0.0, we now have a conda channel: `huggingface`.
-
-🤗 Transformers can be installed using conda as follows:
-
-```
+```bash
 conda install -c huggingface transformers
 ```
 
-Follow the installation pages of TensorFlow, PyTorch or Flax to see how to install them with conda.
+## Cache setup
 
-## Caching models
+Pretrained models are downloaded and locally cached at: `~/.cache/huggingface/transformers/`. This is the default directory given by the shell environment variable `TRANSFORMERS_CACHE`. On Windows, the default directory is given by `C:\Users\username\.cache\huggingface\transformers`. You can change the shell environment variables shown below - in order of priority - to specify a different cache directory:
 
-This library provides pretrained models that will be downloaded and cached locally. Unless you specify a location with
-`cache_dir=...` when you use methods like `from_pretrained`, these models will automatically be downloaded in the
-folder given by the shell environment variable ``TRANSFORMERS_CACHE``. The default value for it will be the Hugging
-Face cache home followed by ``/transformers/``. This is (by order of priority):
+1. Shell environment variable (default): `TRANSFORMERS_CACHE`.
+2. Shell environment variable: `HF_HOME` + `transformers/`.
+3. Shell environment variable: `XDG_CACHE_HOME` + `/huggingface/transformers`.
 
-  * shell environment variable ``HF_HOME``
-  * shell environment variable ``XDG_CACHE_HOME`` + ``/huggingface/``
-  * default: ``~/.cache/huggingface/``
+<Tip>
 
-So if you don't have any specific environment variable set, the cache directory will be at
-``~/.cache/huggingface/transformers/``.
+🤗 Transformers will use the shell environment variables `PYTORCH_TRANSFORMERS_CACHE` or `PYTORCH_PRETRAINED_BERT_CACHE` if you are coming from an earlier iteration of this library and have set those environment variables, unless you specify the shell environment variable `TRANSFORMERS_CACHE`.
 
-**Note:** If you have set a shell environment variable for one of the predecessors of this library
-(``PYTORCH_TRANSFORMERS_CACHE`` or ``PYTORCH_PRETRAINED_BERT_CACHE``), those will be used if there is no shell
-environment variable for ``TRANSFORMERS_CACHE``.
+</Tip>
 
-### Offline mode
+## Offline mode
 
-It's possible to run 🤗 Transformers in a firewalled or a no-network environment.
+🤗 Transformers is able to run in a firewalled or offline environment by only using local files. Set the environment variable `TRANSFORMERS_OFFLINE=1` to enable this behavior.
 
-Setting environment variable `TRANSFORMERS_OFFLINE=1` will tell 🤗 Transformers to use local files only and will not try to look things up.
+<Tip>
 
-Most likely you may want to couple this with `HF_DATASETS_OFFLINE=1` that performs the same for 🤗 Datasets if you're using the latter.
+Add [🤗 Datasets](https://huggingface.co/docs/datasets/) to your offline training workflow by setting the environment variable `HF_DATASETS_OFFLINE=1`.
 
-Here is an example of how this can be used on a filesystem that is shared between a normally networked and a firewalled to the external world instances.
+</Tip>
 
-On the instance with the normal network run your program which will download and cache models (and optionally datasets if you use 🤗 Datasets). For example:
+For example, you would typically run a program on a normal network firewalled to external instances with the following command:
 
-```
+```bash
 python examples/pytorch/translation/run_translation.py --model_name_or_path t5-small --dataset_name wmt16 --dataset_config ro-en ...
 ```
 
-and then with the same filesystem you can now run the same program on a firewalled instance:
-```
+Run this same program in an offline instance with:
+
+```bash
 HF_DATASETS_OFFLINE=1 TRANSFORMERS_OFFLINE=1 \
 python examples/pytorch/translation/run_translation.py --model_name_or_path t5-small --dataset_name wmt16 --dataset_config ro-en ...
 ```
-and it should succeed without any hanging waiting to timeout.
 
-#### Fetching models and tokenizers to use offline
+The script should now run without hanging or waiting to timeout because it knows it should only look for local files.
 
-When running a script the first time like mentioned above, the downloaded files will be cached for future reuse. 
-However, it is also possible to download files and point to their local path instead.
+### Fetch models and tokenizers to use offline
 
-Downloading files can be done through the Web Interface by clicking on the "Download" button, but it can also be handled
-programmatically using the `huggingface_hub` library that is a dependency to `transformers`:
+Another option for using 🤗 Transformers offline is to download the files ahead of time, and then point to their local path when you need to use them offline. There are three ways to do this:
 
-- Using `snapshot_download` to download an entire repository
-- Using `hf_hub_download` to download a specific file
+* Download a file through the user interface on the [Model Hub](https://huggingface.co/models) by clicking on the ↓ icon.
 
-See the reference for these methods in the huggingface_hub
-[documentation](https://github.com/huggingface/huggingface_hub/tree/main/src/huggingface_hub).
+    ![download-icon](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/download-icon.png)
 
-## Do you want to run a Transformer model on a mobile device?
+* Use the [`PreTrainedModel.from_pretrained`] and [`PreTrainedModel.save_pretrained`] workflow:
 
-You should check out our [swift-coreml-transformers](https://github.com/huggingface/swift-coreml-transformers) repo.
+    1. Download your files ahead of time with [`PreTrainedModel.from_pretrained`]:
 
-It contains a set of tools to convert PyTorch or TensorFlow 2.0 trained Transformer models (currently contains `GPT-2`,
-`DistilGPT-2`, `BERT`, and `DistilBERT`) to CoreML models that run on iOS devices.
+    ```py
+    >>> from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 
-At some point in the future, you'll be able to seamlessly move from pretraining or fine-tuning models in PyTorch or
-TensorFlow 2.0 to productizing them in CoreML, or prototype a model or an app in CoreML then research its
-hyperparameters or architecture from PyTorch or TensorFlow 2.0. Super exciting!
+    >>> tokenizer = AutoTokenizer.from_pretrained("bigscience/T0_3B")
+    >>> model = AutoModelForSeq2SeqLM.from_pretrained("bigscience/T0_3B")
+    ```
+
+    2. Save your files to a specified directory with [`PreTrainedModel.save_pretrained`]:
+
+    ```py
+    >>> tokenizer.save_pretrained("./your/path/bigscience_t0")
+    >>> model.save_pretrained("./your/path/bigscience_t0")
+    ```
+
+    3. Now when you're offline, reload your files with [`PreTrainedModel.from_pretrained`] from the specified directory:
+
+    ```py
+    >>> tokenizer = AutoTokenizer.from_pretrained("./your/path/bigscience_t0")
+    >>> model = AutoModel.from_pretrained("./your/path/bigscience_t0")
+    ```
+
+* Programmatically download files with the [huggingface_hub](https://github.com/huggingface/huggingface_hub/tree/main/src/huggingface_hub) library:
+
+    1. Install the `huggingface_hub` library in your virtual environment:
+
+    ```bash
+    python -m pip install huggingface_hub
+    ```
+
+    2. Use the [`hf_hub_download`](https://huggingface.co/docs/hub/adding-a-library#download-files-from-the-hub) function to download a file to a specific path. For example, the following command downloads the `config.json` file from the [T0](https://huggingface.co/bigscience/T0_3B) model to your desired path:
+
+    ```py
+    >>> from huggingface_hub import hf_hub_download
+
+    >>> hf_hub_download(repo_id="bigscience/T0_3B", filename="config.json", cache_dir="./your/path/bigscience_t0")
+    ```
+
+Once your file is downloaded and locally cached, specify it's local path to load and use it:
+
+```py
+>>> from transformers import AutoConfig
+
+>>> config = AutoConfig.from_pretrained("./your/path/bigscience_t0/config.json")
+```
+
+<Tip>
+
+See the [How to download files from the Hub](https://huggingface.co/docs/hub/how-to-downstream) section for more details on downloading files stored on the Hub.
+
+</Tip>
\ No newline at end of file
diff --git a/docs/source/quicktour.mdx b/docs/source/quicktour.mdx
index 38a65cb528..867170656e 100644
--- a/docs/source/quicktour.mdx
+++ b/docs/source/quicktour.mdx
@@ -1,4 +1,4 @@
-<!--Copyright 2020 The HuggingFace Team. All rights reserved.
+<!--Copyright 2022 The HuggingFace Team. All rights reserved.
 
 Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
 the License. You may obtain a copy of the License at
@@ -14,40 +14,53 @@ specific language governing permissions and limitations under the License.
 
 [[open-in-colab]]
 
-Let's have a quick look at the 🤗 Transformers library features. The library downloads pretrained models for Natural
-Language Understanding (NLU) tasks, such as analyzing the sentiment of a text, and Natural Language Generation (NLG),
-such as completing a prompt with new text or translating in another language.
-
-First we will see how to easily leverage the pipeline API to quickly use those pretrained models at inference. Then, we
-will dig a little bit more and see how the library gives you access to those models and helps you preprocess your data.
+Get up and running with 🤗 Transformers! Start using the [`pipeline`] for rapid inference, and quickly load a pretrained model and tokenizer with an [AutoClass](./model_doc/auto) to solve your text, vision or audio task.
 
 <Tip>
 
-All code examples presented in the documentation have a switch on the top left for Pytorch versus TensorFlow. If
-not, the code is expected to work for both backends without any change needed.
+All code examples presented in the documentation have a toggle on the top left for PyTorch and TensorFlow. If
+not, the code is expected to work for both backends without any change.
 
 </Tip>
 
-## Getting started on a task with a pipeline
+## Pipeline
 
-The easiest way to use a pretrained model on a given task is to use [`~transformers.pipeline`].
+[`pipeline`] is the easiest way to use a pretrained model for a given task.
 
 <Youtube id="tiZFewofSLM"/>
 
-🤗 Transformers provides the following tasks out of the box:
+The [`pipeline`] supports many common tasks out-of-the-box:
 
-- Sentiment analysis: is a text positive or negative?
-- Text generation (in English): provide a prompt and the model will generate what follows.
-- Name entity recognition (NER): in an input sentence, label each word with the entity it represents (person, place, etc.)
-- Question answering: provide the model with some context and a question, extract the answer from the context.
-- Filling masked text: given a text with masked words (e.g., replaced by `[MASK]`), fill the blanks.
-- Summarization: generate a summary of a long text.
-- Translation: translate a text in another language.
-- Feature extraction: return a tensor representation of the text.
+**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.
 
-Let's see how this work for sentiment analysis (the other tasks are all covered in the [task summary](task_summary)):
+**Image**:
+* Image classification: classify an image.
+* Image segmentation: classify every pixel in an image.
+* Object detection: detect objects within an image.
 
-Install the following dependencies (if not already installed):
+**Audio**:
+* Audio classification: assign a label to a given segment of audio.
+* Automatic speech recognition (ASR): transcribe audio data into text.
+
+<Tip>
+
+For more details about the [`pipeline`] and associated tasks, refer to the documentation [here](./main_classes/pipelines).
+
+</Tip>
+
+### Pipeline usage
+
+In the following example, you will use the [`pipeline`] for sentiment analysis.
+
+Install the following dependencies if you haven't already:
 
 ```bash
 pip install torch
@@ -55,25 +68,22 @@ pip install torch
 pip install tensorflow
 ```
 
+Import [`pipeline`] and specify the task you want to complete:
+
 ```py
 >>> from transformers import pipeline
 
 >>> classifier = pipeline("sentiment-analysis")
 ```
 
-When typing this command for the first time, a pretrained model and its tokenizer are downloaded and cached. We will
-look at both later on, but as an introduction the tokenizer's job is to preprocess the text for the model, which is
-then responsible for making predictions. The pipeline groups all of that together, and post-process the predictions to
-make them readable. For instance:
-
+The pipeline downloads and caches a default [pretrained model](https://huggingface.co/distilbert-base-uncased-finetuned-sst-2-english) and tokenizer for sentiment analysis. Now you can use the `classifier` on your target text:
 
 ```py
 >>> classifier("We are very happy to show you the 🤗 Transformers library.")
-[{'label': 'POSITIVE', 'score': 0.9998}]
+[{"label": "POSITIVE", "score": 0.9998}]
 ```
 
-That's encouraging! You can use it on a list of sentences, which will be preprocessed then fed to the model, returning
-a list of dictionaries like this one:
+For more than one sentence, pass a list of sentences to the [`pipeline`] which returns a list of dictionaries:
 
 ```py
 >>> results = classifier(["We are very happy to show you the 🤗 Transformers library.", "We hope you don't hate it."])
@@ -83,117 +93,109 @@ label: POSITIVE, with score: 0.9998
 label: NEGATIVE, with score: 0.5309
 ```
 
-To use with a large dataset, look at [iterating over a pipeline](main_classes/pipelines)
+The [`pipeline`] can also iterate over an entire dataset. Start by installing the [🤗 Datasets](https://huggingface.co/docs/datasets/) library:
 
-You can see the second sentence has been classified as negative (it needs to be positive or negative) but its score is
-fairly neutral.
-
-By default, the model downloaded for this pipeline is called "distilbert-base-uncased-finetuned-sst-2-english". We can
-look at its [model page](https://huggingface.co/distilbert-base-uncased-finetuned-sst-2-english) to get more
-information about it. It uses the [DistilBERT architecture](model_doc/distilbert) and has been fine-tuned on a
-dataset called SST-2 for the sentiment analysis task.
-
-Let's say we want to use another model; for instance, one that has been trained on French data. We can search through
-the [model hub](https://huggingface.co/models) that gathers models pretrained on a lot of data by research labs, but
-also community models (usually fine-tuned versions of those big models on a specific dataset). Applying the tags
-"French" and "text-classification" gives back a suggestion "nlptown/bert-base-multilingual-uncased-sentiment". Let's
-see how we can use it.
-
-You can directly pass the name of the model to use to [`pipeline`]:
-
-```py
->>> classifier = pipeline("sentiment-analysis", model="nlptown/bert-base-multilingual-uncased-sentiment")
+```bash
+pip install datasets 
 ```
 
-This classifier can now deal with texts in English, French, but also Dutch, German, Italian and Spanish! You can also
-replace that name by a local folder where you have saved a pretrained model (see below). You can also pass a model
-object and its associated tokenizer.
+Create a [`pipeline`] with the task you want to solve for and the model you want to use. Set the `device` parameter to `0` to place the tensors on a CUDA device:
 
-We will need two classes for this. The first is [`AutoTokenizer`], which we will use to download the tokenizer associated to the model we picked and instantiate it. The second is [`AutoModelForSequenceClassification`] (or [`TFAutoModelForSequenceClassification`] if you are using TensorFlow), which we will use to download the model itself. Note that if we were using the library on an other task, the class of the model would change. The [task summary](task_summary) tutorial summarizes which class is used for which task.
+```py
+>>> from transformers import pipeline
+
+>>> speech_recognizer = pipeline("automatic-speech-recognition", model="facebook/wav2vec2-base-960h", device=0)
+```
+
+Next, load a dataset (see the 🤗 Datasets [Quick Start](https://huggingface.co/docs/datasets/quickstart.html) for more details) you'd like to iterate over. For example, let's load the [SUPERB](https://huggingface.co/datasets/superb) dataset:
+
+```py
+>>> import datasets
+
+>>> dataset = datasets.load_dataset("superb", name="asr", split="test")
+```
+
+Now you can iterate over the dataset with the pipeline. `KeyDataset` retrieves the item in the dictionary returned by the dataset:
+
+```py
+>>> from transformers.pipelines.base import KeyDataset
+>>> from tqdm.auto import tqdm
+
+>>> for out in tqdm(speech_recognizer(KeyDataset(dataset, "file"))):
+...     print(out)
+{"text": "HE HOPED THERE WOULD BE STEW FOR DINNER TURNIPS AND CARROTS AND BRUISED POTATOES AND FAT MUTTON PIECES TO BE LADLED OUT IN THICK PEPPERED FLOWER FAT AND SAUCE"}
+```
+
+### Use another model and tokenizer in the pipeline
+
+The [`pipeline`] can accommodate any model from the [Model Hub](https://huggingface.co/models), making it easy to adapt the [`pipeline`] for other use-cases. For example, if you'd like a model capable of handling French text, use the tags on the Model Hub to filter for an appropriate model. The top filtered result returns a multilingual [BERT model](https://huggingface.co/nlptown/bert-base-multilingual-uncased-sentiment) fine-tuned for sentiment analysis. Great, let's use this model!
+
+```py
+>>> model_name = "nlptown/bert-base-multilingual-uncased-sentiment"
+```
+
+Use the [`AutoModelForSequenceClassification`] and ['AutoTokenizer'] to load the pretrained model and it's associated tokenizer (more on an `AutoClass` below):
 
 ```py
 >>> from transformers import AutoTokenizer, AutoModelForSequenceClassification
-===PT-TF-SPLIT===
->>> from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
-```
 
-Now, to download the models and tokenizer we found previously, we just have to use the
-[`~transformers.AutoModelForSequenceClassification.from_pretrained`] method (feel free to replace `model_name` by
-any other model from the model hub):
-
-```py
->>> model_name = "nlptown/bert-base-multilingual-uncased-sentiment"
 >>> model = AutoModelForSequenceClassification.from_pretrained(model_name)
 >>> tokenizer = AutoTokenizer.from_pretrained(model_name)
->>> classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)
 ===PT-TF-SPLIT===
->>> model_name = "nlptown/bert-base-multilingual-uncased-sentiment"
->>> # This model only exists in PyTorch, so we use the _from_pt_ flag to import that model in TensorFlow.
->>> model = TFAutoModelForSequenceClassification.from_pretrained(model_name, from_pt=True)
+>>> from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
+
+>>> model = TFAutoModelForSequenceClassification.from_pretrained(model_name)
 >>> tokenizer = AutoTokenizer.from_pretrained(model_name)
->>> classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)
 ```
 
-If you don't find a model that has been pretrained on some data similar to yours, you will need to fine-tune a
-pretrained model on your data. We provide [example scripts](examples) to do so. Once you're done, don't forget
-to share your fine-tuned model on the hub with the community, using [this tutorial](model_sharing).
+Then you can specify the model and tokenizer in the [`pipeline`], and apply the `classifier` on your target text:
 
-<a id='pretrained-model'></a>
+```py
+>>> classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)
+>>> classifier("Nous sommes très heureux de vous présenter la bibliothèque 🤗 Transformers.")
+[{"label": "5 stars", "score": 0.7272651791572571}]
+```
 
-## Under the hood: pretrained models
+If you can't find a model for your use-case, you will need to fine-tune a pretrained model on your data. Take a look at our [fine-tuning tutorial](./training) to learn how. Finally, after you've fine-tuned your pretrained model, please consider sharing it (see tutorial [here](./model_sharing)) with the community on the Model Hub to democratize NLP for everyone! 🤗
 
-Let's now see what happens beneath the hood when using those pipelines.
+## AutoClass
 
 <Youtube id="AhChOFRegn4"/>
 
-As we saw, the model and tokenizer are created using the `from_pretrained` method:
+Under the hood, the [`AutoModelForSequenceClassification`] and [`AutoTokenizer`] classes work together to power the [`pipeline`]. An [AutoClass](./model_doc/auto) is a shortcut that automatically retrieves the architecture of a pretrained model from it's name or path. You only need to select the appropriate `AutoClass` for your task and it's associated tokenizer with [`AutoTokenizer`]. 
+
+Let's return to our example and see how you can use the `AutoClass` to replicate the results of the [`pipeline`].
+
+### AutoTokenizer
+
+A tokenizer is responsible for preprocessing text into a format that is understandable to the model. First, the tokenizer will split the text into words called *tokens*. There are multiple rules that govern the tokenization process, including how to split a word and at what level (learn more about tokenization [here](./tokenizer_summary)). The most important thing to remember though is you need to instantiate the tokenizer with the same model name to ensure you're using the same tokenization rules a model was pretrained with.
+
+Load a tokenizer with [`AutoTokenizer`]:
 
 ```py
->>> from transformers import AutoTokenizer, AutoModelForSequenceClassification
+>>> from transformers import AutoTokenizer
 
->>> model_name = "distilbert-base-uncased-finetuned-sst-2-english"
->>> pt_model = AutoModelForSequenceClassification.from_pretrained(model_name)
->>> tokenizer = AutoTokenizer.from_pretrained(model_name)
-===PT-TF-SPLIT===
->>> from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
-
->>> model_name = "distilbert-base-uncased-finetuned-sst-2-english"
->>> tf_model = TFAutoModelForSequenceClassification.from_pretrained(model_name)
+>>> model_name = "nlptown/bert-base-multilingual-uncased-sentiment"
 >>> tokenizer = AutoTokenizer.from_pretrained(model_name)
 ```
 
-### Using the tokenizer
+Next, the tokenizer converts the tokens into numbers in order to construct a tensor as input to the model. This is known as the model's *vocabulary*.
 
-We mentioned the tokenizer is responsible for the preprocessing of your texts. First, it will split a given text in
-words (or part of words, punctuation symbols, etc.) usually called _tokens_. There are multiple rules that can govern
-that process (you can learn more about them in the [tokenizer summary](tokenizer_summary)), which is why we need
-to instantiate the tokenizer using the name of the model, to make sure we use the same rules as when the model was
-pretrained.
-
-The second step is to convert those _tokens_ into numbers, to be able to build a tensor out of them and feed them to
-the model. To do this, the tokenizer has a _vocab_, which is the part we download when we instantiate it with the
-`from_pretrained` method, since we need to use the same _vocab_ as when the model was pretrained.
-
-To apply these steps on a given text, we can just feed it to our tokenizer:
+Pass your text to the tokenizer:
 
 ```py
->>> inputs = tokenizer("We are very happy to show you the 🤗 Transformers library.")
+>>> encoding = tokenizer("We are very happy to show you the 🤗 Transformers library.")
+>>> print(encoding)
+{"input_ids": [101, 2057, 2024, 2200, 3407, 2000, 2265, 2017, 1996, 100, 19081, 3075, 1012, 102],
+ "attention_mask": [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]}
 ```
 
-This returns a dictionary string to list of ints. It contains the [ids of the tokens](glossary#input-ids), as
-mentioned before, but also additional arguments that will be useful to the model. Here for instance, we also have an
-[attention mask](glossary#attention-mask) that the model will use to have a better understanding of the sequence:
+The tokenizer will return a dictionary containing:
 
+* [input_ids](./glossary#input-ids): numerical representions of your tokens.
+* [atttention_mask](.glossary#attention-mask): indicates which tokens should be attended to.
 
-```py
->>> print(inputs)
-{'input_ids': [101, 2057, 2024, 2200, 3407, 2000, 2265, 2017, 1996, 100, 19081, 3075, 1012, 102],
- 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]}
-```
-
-You can pass a list of sentences directly to your tokenizer. If your goal is to send them through your model as a
-batch, you probably want to pad them all to the same length, truncate them to the maximum length the model can accept
-and get tensors back. You can specify all of that to the tokenizer:
+Just like the [`pipeline`], the tokenizer will accept a list of inputs. In addition, the tokenizer can also pad and truncate the text to return a batch with uniform length:
 
 ```py
 >>> pt_batch = tokenizer(
@@ -213,28 +215,31 @@ and get tensors back. You can specify all of that to the tokenizer:
 ... )
 ```
 
-The padding is automatically applied on the side expected by the model (in this case, on the right), with the padding
-token the model was pretrained with. The attention mask is also adapted to take the padding into account:
+Read the [preprocessing](./preprocessing) tutorial for more details about tokenization.
+
+### AutoModel
+
+🤗 Transformers provides a simple and unified way to load pretrained instances. This means you can load an [`AutoModel`] like you would load an [`AutoTokenizer`]. The only difference is selecting the correct [`AutoModel`] for the task. Since you are doing text - or sequence - classification, load [`AutoModelForSequenceClassification`]. The TensorFlow equivalent is simply [`TFAutoModelForSequenceClassification`]:
 
 ```py
->>> for key, value in pt_batch.items():
-...     print(f"{key}: {value.numpy().tolist()}")
-input_ids: [[101, 2057, 2024, 2200, 3407, 2000, 2265, 2017, 1996, 100, 19081, 3075, 1012, 102], [101, 2057, 3246, 2017, 2123, 1005, 1056, 5223, 2009, 1012, 102, 0, 0, 0]]
-attention_mask: [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0]]
+>>> from transformers import AutoModelForSequenceClassification
+
+>>> model_name = "nlptown/bert-base-multilingual-uncased-sentiment"
+>>> pt_model = AutoModelForSequenceClassification.from_pretrained(model_name)
 ===PT-TF-SPLIT===
->>> for key, value in tf_batch.items():
-...     print(f"{key}: {value.numpy().tolist()}")
-input_ids: [[101, 2057, 2024, 2200, 3407, 2000, 2265, 2017, 1996, 100, 19081, 3075, 1012, 102], [101, 2057, 3246, 2017, 2123, 1005, 1056, 5223, 2009, 1012, 102, 0, 0, 0]]
-attention_mask: [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0]]
+>>> from transformers import TFAutoModelForSequenceClassification
+
+>>> model_name = "nlptown/bert-base-multilingual-uncased-sentiment"
+>>> tf_model = TFAutoModelForSequenceClassification.from_pretrained(model_name)
 ```
 
-You can learn more about tokenizers [here](preprocessing).
+<Tip>
 
-### Using the model
+See the [task summary](./task_summary) for which [`AutoModel`] class to use for which task.
 
-Once your input has been preprocessed by the tokenizer, you can send it directly to the model. As we mentioned, it will
-contain all the relevant information the model needs. If you're using a TensorFlow model, you can pass the dictionary
-keys directly to tensors, for a PyTorch model, you need to unpack the dictionary by adding `**`.
+</Tip>
+
+Now you can pass your preprocessed batch of inputs directly to the model. If you are using a PyTorch model, unpack the dictionary by adding `**`. For TensorFlow models, pass the dictionary keys directly to the tensors:
 
 ```py
 >>> pt_outputs = pt_model(**pt_batch)
@@ -242,85 +247,44 @@ keys directly to tensors, for a PyTorch model, you need to unpack the dictionary
 >>> tf_outputs = tf_model(tf_batch)
 ```
 
-In 🤗 Transformers, all outputs are objects that contain the model's final activations along with other metadata. These
-objects are described in greater detail [here](main_classes/output). For now, let's inspect the output ourselves:
-
-```py
->>> print(pt_outputs)
-SequenceClassifierOutput(loss=None, logits=tensor([[-4.0833,  4.3364],
-        [ 0.0818, -0.0418]], grad_fn=<AddmmBackward>), hidden_states=None, attentions=None)
-===PT-TF-SPLIT===
->>> print(tf_outputs)
-TFSequenceClassifierOutput(loss=None, logits=<tf.Tensor: shape=(2, 2), dtype=float32, numpy=
-array([[-4.0833 ,  4.3364  ],
-       [ 0.0818, -0.0418]], dtype=float32)>, hidden_states=None, attentions=None)
-```
-
-Notice how the output object has a `logits` attribute. You can use this to access the model's final activations.
-
-<Tip>
-
-All 🤗 Transformers models (PyTorch or TensorFlow) return the activations of the model *before* the final activation
-function (like SoftMax) since this final activation function is often fused with the loss.
-
-</Tip>
-
-Let's apply the SoftMax activation to get predictions.
+The model outputs the final activations in the `logits` attribute. Apply the softmax function to the `logits` to retrieve the probabilities:
 
 ```py
 >>> from torch import nn
 
 >>> pt_predictions = nn.functional.softmax(pt_outputs.logits, dim=-1)
-===PT-TF-SPLIT===
->>> import tensorflow as tf
-
->>> tf_predictions = tf.nn.softmax(tf_outputs.logits, axis=-1)
-```
-
-We can see we get the numbers from before:
-
-```py
 >>> print(pt_predictions)
 tensor([[2.2043e-04, 9.9978e-01],
         [5.3086e-01, 4.6914e-01]], grad_fn=<SoftmaxBackward>)
 ===PT-TF-SPLIT===
+>>> import tensorflow as tf
+
+>>> tf_predictions = tf.nn.softmax(tf_outputs.logits, axis=-1)
 >>> print(tf_predictions)
 tf.Tensor(
 [[2.2043e-04 9.9978e-01]
  [5.3086e-01 4.6914e-01]], shape=(2, 2), dtype=float32)
 ```
 
-If you provide the model with labels in addition to inputs, the model output object will also contain a `loss`
-attribute:
-
-```py
->>> import torch
-
->>> pt_outputs = pt_model(**pt_batch, labels=torch.tensor([1, 0]))
->>> print(pt_outputs)
-SequenceClassifierOutput(loss=tensor(0.3167, grad_fn=<NllLossBackward>), logits=tensor([[-4.0833,  4.3364],
-        [ 0.0818, -0.0418]], grad_fn=<AddmmBackward>), hidden_states=None, attentions=None)
-===PT-TF-SPLIT===
->>> import tensorflow as tf
-
->>> tf_outputs = tf_model(tf_batch, labels=tf.constant([1, 0]))
->>> print(tf_outputs)
-TFSequenceClassifierOutput(loss=<tf.Tensor: shape=(2,), dtype=float32, numpy=array([2.2051e-04, 6.3326e-01], dtype=float32)>, logits=<tf.Tensor: shape=(2, 2), dtype=float32, numpy=
-array([[-4.0833 ,  4.3364  ],
-       [ 0.0818, -0.0418]], dtype=float32)>, hidden_states=None, attentions=None)
-```
-
-Models are standard [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) or [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) so you can use them in your usual training loop. 🤗 Transformers also provides a [`Trainer`] class to help with your training in PyTorch (taking care of things such as distributed training, mixed precision, etc.) whereas you can leverage the `fit()` method in Keras. See the [training tutorial](training) for more details.
-
 <Tip>
 
-Pytorch model outputs are special dataclasses so that you can get autocompletion for their attributes in an IDE.
-They also behave like a tuple or a dictionary (e.g., you can index with an integer, a slice or a string) in which
-case the attributes not set (that have `None` values) are ignored.
+All 🤗 Transformers models (PyTorch or TensorFlow) outputs the tensors *before* the final activation
+function (like softmax) because the final activation function is often fused with the loss.
 
 </Tip>
 
-Once your model is fine-tuned, you can save it with its tokenizer in the following way:
+Models are a standard [`torch.nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) or a [`tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model) so you can use them in your usual training loop. However, to make things easier, 🤗 Transformers provides a [`Trainer`] class for PyTorch that adds functionality for distributed training, mixed precision, and more. For TensorFlow, you can use the `fit` method from [Keras](https://keras.io/). Refer to the [training tutorial](./training) for more details.
+
+<Tip>
+
+🤗 Transformers model outputs are special dataclasses so their attributes are autocompleted in an IDE.
+The model outputs also behave like a tuple or a dictionary (e.g., you can index with an integer, a slice or a string) in which case the attributes that are `None` are ignored.
+
+</Tip>
+
+### Save a model
+
+Once your model is fine-tuned, you can save it with its tokenizer using [`PreTrainedModel.save_pretrained`]:
 
 ```py
 >>> pt_save_directory = "./pt_save_pretrained"
@@ -332,113 +296,24 @@ Once your model is fine-tuned, you can save it with its tokenizer in the followi
 >>> tf_model.save_pretrained(tf_save_directory)
 ```
 
-You can then load this model back using the [`AutoModel.from_pretrained`] method by passing the
-directory name instead of the model name. One cool feature of 🤗 Transformers is that you can easily switch between
-PyTorch and TensorFlow: any model saved as before can be loaded back either in PyTorch or TensorFlow.
+When you are ready to use the model again, reload it with [`PreTrainedModel.from_pretrained`]:
 
-
-If you would like to load your saved model in the other framework, first make sure it is installed:
-
-```bash
-pip install torch
+```py
+>>> pt_model = AutoModelForSequenceClassification.from_pretrained("./pt_save_pretrained")
 ===PT-TF-SPLIT===
-pip install tensorflow
+>>> tf_model = TFAutoModelForSequenceClassification.from_pretrained("./tf_save_pretrained")
 ```
 
-Then, use the corresponding Auto class to load it like this:
+One particularly cool 🤗 Transformers feature is the ability to save a model and reload it as either a PyTorch or TensorFlow model. The `from_pt` or `from_tf` parameter can convert the model from one framework to the other:
 
 ```py
 >>> from transformers import AutoModel
 
 >>> tokenizer = AutoTokenizer.from_pretrained(tf_save_directory)
->>> pt_model = AutoModel.from_pretrained(tf_save_directory, from_tf=True)
+>>> pt_model = AutoModelForSequenceClassification.from_pretrained(tf_save_directory, from_tf=True)
 ===PT-TF-SPLIT===
 >>> from transformers import TFAutoModel
 
 >>> tokenizer = AutoTokenizer.from_pretrained(pt_save_directory)
->>> tf_model = TFAutoModel.from_pretrained(pt_save_directory, from_pt=True)
-```
-
-Lastly, you can also ask the model to return all hidden states and all attention weights if you need them:
-
-
-```py
->>> pt_outputs = pt_model(**pt_batch, output_hidden_states=True, output_attentions=True)
->>> all_hidden_states = pt_outputs.hidden_states
->>> all_attentions = pt_outputs.attentions
-===PT-TF-SPLIT===
->>> tf_outputs = tf_model(tf_batch, output_hidden_states=True, output_attentions=True)
->>> all_hidden_states = tf_outputs.hidden_states
->>> all_attentions = tf_outputs.attentions
-```
-
-### Accessing the code
-
-The [`AutoModel`] and [`AutoTokenizer`] classes are just shortcuts that will automatically work with any
-pretrained model. Behind the scenes, the library has one model class per combination of architecture plus class, so the
-code is easy to access and tweak if you need to.
-
-In our previous example, the model was called "distilbert-base-uncased-finetuned-sst-2-english", which means it's using
-the [DistilBERT](model_doc/distilbert) architecture. As [`AutoModelForSequenceClassification`] (or [`TFAutoModelForSequenceClassification`] if you are using TensorFlow) was used, the model automatically created is then a [`DistilBertForSequenceClassification`]. You can look at its documentation for all details relevant to that specific model, or browse the source code. This is how you would
-directly instantiate model and tokenizer without the auto magic:
-
-```py
->>> from transformers import DistilBertTokenizer, DistilBertForSequenceClassification
-
->>> model_name = "distilbert-base-uncased-finetuned-sst-2-english"
->>> model = DistilBertForSequenceClassification.from_pretrained(model_name)
->>> tokenizer = DistilBertTokenizer.from_pretrained(model_name)
-===PT-TF-SPLIT===
->>> from transformers import DistilBertTokenizer, TFDistilBertForSequenceClassification
-
->>> model_name = "distilbert-base-uncased-finetuned-sst-2-english"
->>> model = TFDistilBertForSequenceClassification.from_pretrained(model_name)
->>> tokenizer = DistilBertTokenizer.from_pretrained(model_name)
-```
-
-### Customizing the model
-
-If you want to change how the model itself is built, you can define a custom configuration class. Each architecture
-comes with its own relevant configuration. For example, [`DistilBertConfig`] allows you to specify
-parameters such as the hidden dimension, dropout rate, etc for DistilBERT. If you do core modifications, like changing
-the hidden size, you won't be able to use a pretrained model anymore and will need to train from scratch. You would
-then instantiate the model directly from this configuration.
-
-Below, we load a predefined vocabulary for a tokenizer with the
-[`~DistilBertTokenizer.from_pretrained`] method. However, unlike the tokenizer, we wish to initialize
-the model from scratch. Therefore, we instantiate the model from a configuration instead of using the
-[`DistilBertForSequenceClassification.from_pretrained`] method.
-
-```py
->>> from transformers import DistilBertConfig, DistilBertTokenizer, DistilBertForSequenceClassification
-
->>> config = DistilBertConfig(n_heads=8, dim=512, hidden_dim=4 * 512)
->>> tokenizer = DistilBertTokenizer.from_pretrained("distilbert-base-uncased")
->>> model = DistilBertForSequenceClassification(config)
-===PT-TF-SPLIT===
->>> from transformers import DistilBertConfig, DistilBertTokenizer, TFDistilBertForSequenceClassification
-
->>> config = DistilBertConfig(n_heads=8, dim=512, hidden_dim=4 * 512)
->>> tokenizer = DistilBertTokenizer.from_pretrained("distilbert-base-uncased")
->>> model = TFDistilBertForSequenceClassification(config)
-```
-
-For something that only changes the head of the model (for instance, the number of labels), you can still use a
-pretrained model for the body. For instance, let's define a classifier for 10 different labels using a pretrained body.
-Instead of creating a new configuration with all the default values just to change the number of labels, we can instead
-pass any argument a configuration would take to the [`from_pretrained`] method and it will update the default
-configuration appropriately:
-
-```py
->>> from transformers import DistilBertConfig, DistilBertTokenizer, DistilBertForSequenceClassification
-
->>> model_name = "distilbert-base-uncased"
->>> model = DistilBertForSequenceClassification.from_pretrained(model_name, num_labels=10)
->>> tokenizer = DistilBertTokenizer.from_pretrained(model_name)
-===PT-TF-SPLIT===
->>> from transformers import DistilBertConfig, DistilBertTokenizer, TFDistilBertForSequenceClassification
-
->>> model_name = "distilbert-base-uncased"
->>> model = TFDistilBertForSequenceClassification.from_pretrained(model_name, num_labels=10)
->>> tokenizer = DistilBertTokenizer.from_pretrained(model_name)
-```
+>>> tf_model = TFAutoModelForSequenceClassification.from_pretrained(pt_save_directory, from_pt=True)
+```
\ No newline at end of file