Phuc Van Phan 910faa3e1f feat: adding num_proc to load_dataset (#26326) 1 day ago
flax 910faa3e1f feat: adding num_proc to load_dataset (#26326) 1 day ago
legacy 404ff8fc17 Fix typo (#25966) 2 weeks ago
pytorch 4fb64e285a chore: correct update_step and correct gradient_accumulation_steps (#26068) 1 week ago
research_projects 5af2c62696 docs: add space to docs (#26067) 1 week ago
tensorflow 7d4e0c23c8 Update (#26198) 5 days ago f6295c6c53 Update (#26003) 2 weeks ago fd8dcd0953 Udate link to RunHouse hardware setup documentation. (#24590) 2 months ago


We host a wide range of example scripts for multiple learning frameworks. Simply choose your favorite: TensorFlow, PyTorch or JAX/Flax.

We also have some research projects, as well as some legacy examples. Note that unlike the main examples these are not actively maintained, and may require specific older versions of dependencies in order to run.

While we strive to present as many use cases as possible, the example scripts are just that - examples. It is expected that they won't work out-of-the-box on your specific problem and that you will be required to change a few lines of code to adapt them to your needs. To help you with that, most of the examples fully expose the preprocessing of the data, allowing you to tweak and edit them as required.

Please discuss on the forum or in an issue a feature you would like to implement in an example before submitting a PR; we welcome bug fixes, but since we want to keep the examples as simple as possible it's unlikely that we will merge a pull request adding more functionality at the cost of readability.

Important note


To make sure you can successfully run the latest versions of the example scripts, you have to install the library from source and install some example-specific requirements. To do this, execute the following steps in a new virtual environment:

git clone
cd transformers
pip install .

Then cd in the example folder of your choice and run

pip install -r requirements.txt

To browse the examples corresponding to released versions of 🤗 Transformers, click on the line below and then on your desired version of the library:

Examples for older versions of 🤗 Transformers
    <li><a href="">v4.21.0</a></li>
    <li><a href="">v4.20.1</a></li>
    <li><a href="">v4.19.4</a></li>
    <li><a href="">v4.18.0</a></li>
    <li><a href="">v4.17.0</a></li>
    <li><a href="">v4.16.2</a></li>
    <li><a href="">v4.15.0</a></li>
    <li><a href="">v4.14.1</a></li>
    <li><a href="">v4.13.0</a></li>
    <li><a href="">v4.12.5</a></li>
    <li><a href="">v4.11.3</a></li>
    <li><a href="">v4.10.3</a></li>
    <li><a href="">v4.9.2</a></li>
    <li><a href="">v4.8.2</a></li>
    <li><a href="">v4.7.0</a></li>
    <li><a href="">v4.6.1</a></li>
    <li><a href="">v4.5.1</a></li>
    <li><a href="">v4.4.2</a></li>
    <li><a href="">v4.3.3</a></li>
    <li><a href="">v4.2.2</a></li>
    <li><a href="">v4.1.1</a></li>
    <li><a href="">v4.0.1</a></li>
    <li><a href="">v3.5.1</a></li>
    <li><a href="">v3.4.0</a></li>
    <li><a href="">v3.3.1</a></li>
    <li><a href="">v3.2.0</a></li>
    <li><a href="">v3.1.0</a></li>
    <li><a href="">v3.0.2</a></li>
    <li><a href="">v2.11.0</a></li>
    <li><a href="">v2.10.0</a></li>
    <li><a href="">v2.9.1</a></li>
    <li><a href="">v2.8.0</a></li>
    <li><a href="">v2.7.0</a></li>
    <li><a href="">v2.6.0</a></li>
    <li><a href="">v2.5.1</a></li>
    <li><a href="">v2.4.0</a></li>
    <li><a href="">v2.3.0</a></li>
    <li><a href="">v2.2.0</a></li>
    <li><a href="">v2.1.1</a></li>
    <li><a href="">v2.0.0</a></li>
    <li><a href="">v1.2.0</a></li>
    <li><a href="">v1.1.0</a></li>
    <li><a href="">v1.0.0</a></li>

Alternatively, you can switch your cloned 🤗 Transformers to a specific version (for instance with v3.5.1) with

git checkout tags/v3.5.1

and run the example command as usual afterward.

Running the Examples on Remote Hardware with Auto-Setup is a script that launches any example on remote self-hosted hardware, with automatic hardware and environment setup. It uses Runhouse to launch on self-hosted hardware (e.g. in your own cloud account or on-premise cluster) but there are other options for running remotely as well. You can easily customize the example used, command line arguments, dependencies, and type of compute hardware, and then run the script to automatically launch the example.

You can refer to hardware setup for more information about hardware and dependency setup with Runhouse, or this Colab tutorial for a more in-depth walkthrough.

You can run the script with the following commands:

# First install runhouse:
pip install runhouse

# For an on-demand V100 with whichever cloud provider you have configured:
python \
    --example pytorch/text-generation/ \
    --model_type=gpt2 \
    --model_name_or_path=gpt2 \
    --prompt "I am a language model and"

# For byo (bring your own) cluster:
python --host <cluster_ip> --user <ssh_user> --key_path <ssh_key_path> \
  --example <example> <args>

# For on-demand instances
python --instance <instance> --provider <provider> \
  --example <example> <args>

You can also adapt the script to your own needs.