As models get bigger, parallelism has emerged as a strategy for training larger models on limited hardware and accelerating training speed by several orders of magnitude. At Hugging Face, we created the 🤗 Accelerate library to help users easily train a 🤗 Transformers model on any type of distributed setup, whether it is multiple GPU’s on one machine or multiple GPU’s across several machines. In this tutorial, learn how to customize your native PyTorch training loop to enable training in a distributed environment.
Get started by installing 🤗 Accelerate:
pip install accelerate
Then import and create an
Accelerator will automatically detect your type of distributed setup and initialize all the necessary components for training. You don’t need to explicitly place your model on a device.
from accelerate import Accelerator accelerator = Accelerator()
The next step is to pass all the relevant training objects to
prepare. This includes your training and evaluation DataLoaders, a model and an optimizer:
train_dataloader, eval_dataloader, model, optimizer = accelerator.prepare( train_dataloader, eval_dataloader, model, optimizer )
The last addition is to replace the typical
loss.backward() in your training loop with 🤗 Accelerate’s
for epoch in range(num_epochs): for batch in train_dataloader: outputs = model(**batch) loss = outputs.loss accelerator.backward(loss) optimizer.step() lr_scheduler.step() optimizer.zero_grad() progress_bar.update(1)
As you can see in the following image, you only need to add four additional lines of code to your training loop to enable distributed training!
Once you’ve added the relevant lines of code, launch your training in a script or a notebook like Colaboratory.
If you are running your training from a script, run the following command to create and save a configuration file:
Then launch your training with:
accelerate launch train.py
🤗 Accelerate can also run in a notebook if you’re planning on using Colaboratory’s TPUs. Wrap all the code responsible for training in a function, and pass it to
from accelerate import notebook_launcher notebook_launcher(training_function)
For more information about 🤗 Accelerate and it’s rich features, refer to the documentation.