Skip to main content

Hugging Face Accelerate

Accelerate is a library that enables the same PyTorch code to be run across any distributed configuration by adding just four lines of code, making training and inference at scale made simple, efficient and adaptable.

Accelerate includes a Weights & Biases Tracker which we show how to use below. You can also read more about Accelerate Trackers in their docs here

Start logging with Accelerate

To get started with Accelerate and Weights & Biases you can follow the pseudocode below:

from accelerate import Accelerator

# Tell the Accelerator object to log with wandb
accelerator = Accelerator(log_with="wandb")

# Initialise your wandb run, passing wandb parameters and any config information
config={"dropout": 0.1, "learning_rate": 1e-2}
init_kwargs={"wandb": {"entity": "my-wandb-team"}}


# Log to wandb by calling `accelerator.log`, `step` is optional
accelerator.log({"train_loss": 1.12, "valid_loss": 0.8}, step=global_step)

# Make sure that the wandb tracker finishes correctly

Explaining more, you need to:

  1. Pass log_with="wandb" when initialising the Accelerator class
  2. Call the init_trackers method and pass it:
  • a project name via project_name
  • any parameters you want to pass to wandb.init via a nested dict to init_kwargs
  • any other experiment config information you want to log to your wandb run, via config
  1. Use the .log method to log to Weigths & Biases; the step argument is optional
  2. Call .end_training when finished training

Accessing Accelerates' Internal W&B Tracker

You can quickly access the wandb tracker using the Accelerator.get_tracker() method. Just pass in the string corresponding to a tracker’s .name attribute and it will return that tracker on the main process.

wandb_tracker = accelerator.get_tracker("wandb")

From there you can interact with wandb’s run object like normal:


Trackers built in Accelerate will automatically execute on the correct process, so if a tracker is only meant to be ran on the main process it will do so automatically.

If you want to truly remove Accelerate’s wrapping entirely, you can achieve the same outcome with:

wandb_tracker = accelerator.get_tracker("wandb", unwrap=True)
with accelerator.on_main_process:

Accelerate Articles

Below is an Accelerate article you may enjoy

HuggingFace Accelerate Super Charged With Weights & Biases
  • In this article, we'll look at what HuggingFace Accelerate has to offer and how simple it is to perform distributed training and evaluation, while logging results to Weights & Biases

Read the full report here.

Was this page helpful?👍👎