đŸȘŽ Anil's Garden

Hugging Face Transformers Weights & Biases Documentation

02 Apr 202511 min read

Hugging Face Transformers | Weights & Biases Documentation

Excerpt

The Hugging Face Transformers library makes state-of-the-art NLP models like BERT and training techniques like mixed precision and gradient checkpointing easy to use. The W&B integration adds rich, flexible experiment tracking and model versioning to interactive centralized dashboards without compromising that ease of use.

The Hugging Face Transformers library makes state-of-the-art NLP models like BERT and training techniques like mixed precision and gradient checkpointing easy to use. The W&B integration adds rich, flexible experiment tracking and model versioning to interactive centralized dashboards without compromising that ease of use.

đŸ€— Next-level logging in few lines

<span><span>os</span><span>.</span><span>environ</span><span>[</span><span>"WANDB_PROJECT"</span><span>]</span><span> </span><span>=</span><span> </span><span>"&lt;my-amazing-project&gt;"</span><span>  </span><span># name your W&amp;B project</span><span></span><br></span><span><span>os</span><span>.</span><span>environ</span><span>[</span><span>"WANDB_LOG_MODEL"</span><span>]</span><span> </span><span>=</span><span> </span><span>"checkpoint"</span><span>  </span><span># log all model checkpoints</span><span></span><br></span><span><span></span><br></span><span><span></span><span>from</span><span> transformers </span><span>import</span><span> TrainingArguments</span><span>,</span><span> Trainer</span><br></span><span><span></span><br></span><span><span>args </span><span>=</span><span> TrainingArguments</span><span>(</span><span>.</span><span>.</span><span>.</span><span>,</span><span> report_to</span><span>=</span><span>"wandb"</span><span>)</span><span>  </span><span># turn on W&amp;B logging</span><span></span><br></span><span><span>trainer </span><span>=</span><span> Trainer</span><span>(</span><span>.</span><span>.</span><span>.</span><span>,</span><span> args</span><span>=</span><span>args</span><span>)</span><br></span>

Explore your experiment results in the W&B interactive dashboard

info

If you’d rather dive straight into working code, check out this Google Colab.

Getting started: track experiments

1) Sign Up, install the wandb library and log in

a) Sign up for a free account

b) Pip install the wandb library

c) To log in in your training script, you’ll need to be signed in to you account at www.wandb.ai, then you will find your API key on the Authorize page.

If you are using Weights and Biases for the first time you might want to check out our quickstart

  • Python
  • Command Line
<span><span>pip install wandb</span><br></span><span><span></span><br></span><span><span>wandb login</span><br></span>

2) Name the project

A Project is where all of the charts, data, and models logged from related runs are stored. Naming your project helps you organize your work and keep all the information about a single project in one place.

To add a run to a project simply set the WANDB_PROJECT environment variable to the name of your project. The WandbCallback will pick up this project name environment variable and use it when setting up your run.

  • Python
  • Command Line
  • Notebook
<span><span>import os</span><br></span><span><span>os.environ["WANDB_PROJECT"]="amazon_sentiment_analysis"</span><br></span>

info

Make sure you set the project name before you initialize the Trainer.

If a project name is not specified the project name defaults to “huggingface”.

3) Log your training runs to W&B

This is the most important step: when defining your Trainer training arguments, either inside your code or from the command line, is to set report_to to "wandb" in order enable logging with Weights & Biases.

The logging_steps argument in TrainingArguments will control how often training metrics are pushed to W&B during training. You can also give a name to the training run in W&B using the run_name argument.

That’s it! Now your models will log losses, evaluation metrics, model topology, and gradients to Weights & Biases while they train.

  • Python
  • Command Line
<span><span>from</span><span> transformers </span><span>import</span><span> TrainingArguments</span><span>,</span><span> Trainer</span><br></span><span><span></span><br></span><span><span>args </span><span>=</span><span> TrainingArguments</span><span>(</span><span></span><br></span><span><span>    </span><span># other args and kwargs here</span><span></span><br></span><span><span>    report_to</span><span>=</span><span>"wandb"</span><span>,</span><span>  </span><span># enable logging to W&amp;B</span><span></span><br></span><span><span>    run_name</span><span>=</span><span>"bert-base-high-lr"</span><span>,</span><span>  </span><span># name of the W&amp;B run (optional)</span><span></span><br></span><span><span>    logging_steps</span><span>=</span><span>1</span><span>,</span><span>  </span><span># how often to log to W&amp;B</span><span></span><br></span><span><span></span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>trainer </span><span>=</span><span> Trainer</span><span>(</span><span></span><br></span><span><span>    </span><span># other args and kwargs here</span><span></span><br></span><span><span>    args</span><span>=</span><span>args</span><span>,</span><span>  </span><span># your training args</span><span></span><br></span><span><span></span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>trainer</span><span>.</span><span>train</span><span>(</span><span>)</span><span>  </span><span># start training and logging to W&amp;B</span><br></span>

info

Using TensorFlow? Just swap the PyTorch Trainer for the TensorFlow TFTrainer.

4) Turn on model checkpointing

Using Weights & Biases’ Artifacts, you can store up to 100GB of models and datasets for free and then use the Weights & Biases Model Registry to register models to prepare them for staging or deployment in your production environment.

Logging your Hugging Face model checkpoints to Artifacts can be done by setting the WANDB_LOG_MODEL environment variable to one of end or checkpoint or false:

  • checkpoint: a checkpoint will be uploaded every args.save_steps from the TrainingArguments.
  • end: the model will be uploaded at the end of training.

Use WANDB_LOG_MODEL along with load_best_model_at_end to upload the best model at the end of training.

  • Python
  • Command Line
  • Notebook
<span><span>import</span><span> os</span><br></span><span><span></span><br></span><span><span>os</span><span>.</span><span>environ</span><span>[</span><span>"WANDB_LOG_MODEL"</span><span>]</span><span> </span><span>=</span><span> </span><span>"checkpoint"</span><br></span>

Any Transformers Trainer you initialize from now on will upload models to your W&B project. The model checkpoints you log will be viewable through the Artifacts UI, and include the full model lineage (see an example model checkpoint in the UI here.

info

By default, your model will be saved to W&B Artifacts as model-{run_id} when WANDB_LOG_MODEL is set to end or checkpoint-{run_id} when WANDB_LOG_MODEL is set to checkpoint. However, If you pass a run_name in your TrainingArguments, the model will be saved as model-{run_name} or checkpoint-{run_name}.

W&B Model Registry

Once you have logged your checkpoints to Artifacts, you can then register your best model checkpoints and centralize them across your team using the Weights & Biases Model Registry. Here you can organize your best models by task, manage model lifecycle, facilitate easy tracking and auditing throughout the ML lifecyle, and automate downstream actions with webhooks or jobs.

See the Model Registry documentation for how to link a model Artifact to the Model Registry.

5) Visualise evaluation outputs during training

Visualing your model outputs during training or evaluation is often essential to really understand how your model is training.

By using the callbacks system in the Transformers Trainer, you can log additional helpful data to W&B such as your models’ text generation outputs or other predictions to W&B Tables.

See the Custom logging section below for a full guide on how to log evaluation outupts while training to log to a W&B Table like this:

Shows a W&B Table with evaluation outputs

6) Finish your W&B Run (Notebook only)

If your training is encapsulated in a Python script, the W&B run will end when your script finishes.

If you are using a Jupyter or Google Colab notebook, you’ll need to tell us when you’re done with training by calling wandb.finish().

<span><span>trainer</span><span>.</span><span>train</span><span>(</span><span>)</span><span>  </span><span># start training and logging to W&amp;B</span><span></span><br></span><span><span></span><br></span><span><span></span><span># post-training analysis, testing, other logged code</span><span></span><br></span><span><span></span><br></span><span><span>wandb</span><span>.</span><span>finish</span><span>(</span><span>)</span><br></span>

7) Visualize your results

Once you have logged your training results you can explore your results dynamically in the W&B Dashboard. It’s easy to compare across dozens of runs at once, zoom in on interesting findings, and coax insights out of complex data with flexible, interactive visualizations.

Advanced features and FAQs

How do I save the best model?

If load_best_model_at_end=True is set in the TrainingArguments that are passed to the Trainer, then W&B will save the best performing model checkpoint to Artifacts.

If you’d like to centralize all your best model versions across your team to organize them by ML task, stage them for production, bookmark them for further evaluation, or kick off downstream Model CI/CD processes then ensure you’re saving your model checkpoints to Artifacts. Once logged to Artifacts, these checkpoints can then be promoted to the Model Registry.

Loading a saved model

If you saved your model to W&B Artifacts with WANDB_LOG_MODEL, you can download your model weights for additional training or to run inference. You just load them back into the same Hugging Face architecture that you used before.

<span><span># Create a new run</span><span></span><br></span><span><span></span><span>with</span><span> wandb</span><span>.</span><span>init</span><span>(</span><span>project</span><span>=</span><span>"amazon_sentiment_analysis"</span><span>)</span><span> </span><span>as</span><span> run</span><span>:</span><span></span><br></span><span><span>    </span><span># Pass the name and version of Artifact</span><span></span><br></span><span><span>    my_model_name </span><span>=</span><span> </span><span>"model-bert-base-high-lr:latest"</span><span></span><br></span><span><span>    my_model_artifact </span><span>=</span><span> run</span><span>.</span><span>use_artifact</span><span>(</span><span>my_model_name</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>    </span><span># Download model weights to a folder and return the path</span><span></span><br></span><span><span>    model_dir </span><span>=</span><span> my_model_artifact</span><span>.</span><span>download</span><span>(</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>    </span><span># Load your Hugging Face model from that folder</span><span></span><br></span><span><span>    </span><span>#  using the same model class</span><span></span><br></span><span><span>    model </span><span>=</span><span> AutoModelForSequenceClassification</span><span>.</span><span>from_pretrained</span><span>(</span><span></span><br></span><span><span>        model_dir</span><span>,</span><span> num_labels</span><span>=</span><span>num_labels</span><br></span><span><span>    </span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>    </span><span># Do additional training, or run inference</span><br></span>

Resume training from a checkpoint

If you had set WANDB_LOG_MODEL='checkpoint' you can also resume training by you can using the model_dir as the model_name_or_path argument in your TrainingArguments and pass resume_from_checkpoint=True to Trainer.

<span><span>last_run_id </span><span>=</span><span> </span><span>"xxxxxxxx"</span><span>  </span><span># fetch the run_id from your wandb workspace</span><span></span><br></span><span><span></span><br></span><span><span></span><span># resume the wandb run from the run_id</span><span></span><br></span><span><span></span><span>with</span><span> wandb</span><span>.</span><span>init</span><span>(</span><span></span><br></span><span><span>    project</span><span>=</span><span>os</span><span>.</span><span>environ</span><span>[</span><span>"WANDB_PROJECT"</span><span>]</span><span>,</span><span></span><br></span><span><span>    </span><span>id</span><span>=</span><span>last_run_id</span><span>,</span><span></span><br></span><span><span>    resume</span><span>=</span><span>"must"</span><span>,</span><span></span><br></span><span><span></span><span>)</span><span> </span><span>as</span><span> run</span><span>:</span><span></span><br></span><span><span>    </span><span># Connect an Artifact to the run</span><span></span><br></span><span><span>    my_checkpoint_name </span><span>=</span><span> </span><span>f"checkpoint-</span><span>{</span><span>last_run_id</span><span>}</span><span>:latest"</span><span></span><br></span><span><span>    my_checkpoint_artifact </span><span>=</span><span> run</span><span>.</span><span>use_artifact</span><span>(</span><span>my_model_name</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>    </span><span># Download checkpoint to a folder and return the path</span><span></span><br></span><span><span>    checkpoint_dir </span><span>=</span><span> my_checkpoint_artifact</span><span>.</span><span>download</span><span>(</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>    </span><span># reinitialize your model and trainer</span><span></span><br></span><span><span>    model </span><span>=</span><span> AutoModelForSequenceClassification</span><span>.</span><span>from_pretrained</span><span>(</span><span></span><br></span><span><span>        </span><span>"&lt;model_name&gt;"</span><span>,</span><span> num_labels</span><span>=</span><span>num_labels</span><br></span><span><span>    </span><span>)</span><span></span><br></span><span><span>    </span><span># your awesome training arguments here.</span><span></span><br></span><span><span>    training_args </span><span>=</span><span> TrainingArguments</span><span>(</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>    trainer </span><span>=</span><span> Trainer</span><span>(</span><span>model</span><span>=</span><span>model</span><span>,</span><span> args</span><span>=</span><span>training_args</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span>    </span><span># make sure use the checkpoint dir to resume training from the checkpoint</span><span></span><br></span><span><span>    trainer</span><span>.</span><span>train</span><span>(</span><span>resume_from_checkpoint</span><span>=</span><span>checkpoint_dir</span><span>)</span><br></span>

Custom logging: log and view evaluation samples during training

Logging to Weights & Biases via the Transformers Trainer is taken care of by the WandbCallback in the Transformers library. If you need to customize your Hugging Face logging you can modify this callback by subclassing WandbCallback and adding additional functionality that leverages additional methods from the Trainer class.

Below is the general pattern to add this new callback to the HF Trainer, and further down is a code-complete example to log evaluation outputs to a W&B Table:

<span><span># Instantiate the Trainer as normal</span><span></span><br></span><span><span>trainer </span><span>=</span><span> Trainer</span><span>(</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span></span><span># Instantiate the new logging callback, passing it the Trainer object</span><span></span><br></span><span><span>evals_callback </span><span>=</span><span> WandbEvalsCallback</span><span>(</span><span>trainer</span><span>,</span><span> tokenizer</span><span>,</span><span> </span><span>.</span><span>.</span><span>.</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span></span><span># Add the callback to the Trainer</span><span></span><br></span><span><span>trainer</span><span>.</span><span>add_callback</span><span>(</span><span>evals_callback</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span></span><span># Begin Trainer training as normal</span><span></span><br></span><span><span>trainer</span><span>.</span><span>train</span><span>(</span><span>)</span><br></span>

View evaluation samples during training

The following section shows how to customize the WandbCallback to run model predictions and log evaluation samples to a W&B Table during training. We will every eval_steps using the on_evaluate method of the Trainer callback.

Here, we wrote a decode_predictions function to decode the predictions and labels from the model output using the tokenizer.

Then, we create a pandas DataFrame from the predictions and labels and add an epoch column to the DataFrame.

Finally, we create a wandb.Table from the DataFrame and log it to wandb. Additionally, we can control the frequency of logging by logging the predictions every freq epochs.

Note: Unlike the regular WandbCallback this custom callback needs to be added to the trainer after the Trainer is instantiated and not during initialization of the Trainer. This is because the Trainer instance is passed to the callback during initialization.

<span><span>from</span><span> transformers</span><span>.</span><span>integrations </span><span>import</span><span> WandbCallback</span><br></span><span><span></span><span>import</span><span> pandas </span><span>as</span><span> pd</span><br></span><span><span></span><br></span><span><span></span><br></span><span><span></span><span>def</span><span> </span><span>decode_predictions</span><span>(</span><span>tokenizer</span><span>,</span><span> predictions</span><span>)</span><span>:</span><span></span><br></span><span><span>    labels </span><span>=</span><span> tokenizer</span><span>.</span><span>batch_decode</span><span>(</span><span>predictions</span><span>.</span><span>label_ids</span><span>)</span><span></span><br></span><span><span>    logits </span><span>=</span><span> predictions</span><span>.</span><span>predictions</span><span>.</span><span>argmax</span><span>(</span><span>axis</span><span>=</span><span>-</span><span>1</span><span>)</span><span></span><br></span><span><span>    prediction_text </span><span>=</span><span> tokenizer</span><span>.</span><span>batch_decode</span><span>(</span><span>logits</span><span>)</span><span></span><br></span><span><span>    </span><span>return</span><span> </span><span>{</span><span>"labels"</span><span>:</span><span> labels</span><span>,</span><span> </span><span>"predictions"</span><span>:</span><span> prediction_text</span><span>}</span><span></span><br></span><span><span></span><br></span><span><span></span><br></span><span><span></span><span>class</span><span> </span><span>WandbPredictionProgressCallback</span><span>(</span><span>WandbCallback</span><span>)</span><span>:</span><span></span><br></span><span><span>    </span><span>"""Custom WandbCallback to log model predictions during training.</span><br></span><span><span></span><br></span><span><span>    This callback logs model predictions and labels to a wandb.Table at each </span><br></span><span><span>    logging step during training. It allows to visualize the </span><br></span><span><span>    model predictions as the training progresses.</span><br></span><span><span></span><br></span><span><span>    Attributes:</span><br></span><span><span>        trainer (Trainer): The Hugging Face Trainer instance.</span><br></span><span><span>        tokenizer (AutoTokenizer): The tokenizer associated with the model.</span><br></span><span><span>        sample_dataset (Dataset): A subset of the validation dataset </span><br></span><span><span>          for generating predictions.</span><br></span><span><span>        num_samples (int, optional): Number of samples to select from </span><br></span><span><span>          the validation dataset for generating predictions. Defaults to 100.</span><br></span><span><span>        freq (int, optional): Frequency of logging. Defaults to 2.</span><br></span><span><span>    """</span><span></span><br></span><span><span></span><br></span><span><span>    </span><span>def</span><span> </span><span>__init__</span><span>(</span><span>self</span><span>,</span><span> trainer</span><span>,</span><span> tokenizer</span><span>,</span><span> val_dataset</span><span>,</span><span></span><br></span><span><span>                 num_samples</span><span>=</span><span>100</span><span>,</span><span> freq</span><span>=</span><span>2</span><span>)</span><span>:</span><span></span><br></span><span><span>        </span><span>"""Initializes the WandbPredictionProgressCallback instance.</span><br></span><span><span></span><br></span><span><span>        Args:</span><br></span><span><span>            trainer (Trainer): The Hugging Face Trainer instance.</span><br></span><span><span>            tokenizer (AutoTokenizer): The tokenizer associated </span><br></span><span><span>              with the model.</span><br></span><span><span>            val_dataset (Dataset): The validation dataset.</span><br></span><span><span>            num_samples (int, optional): Number of samples to select from </span><br></span><span><span>              the validation dataset for generating predictions.</span><br></span><span><span>              Defaults to 100.</span><br></span><span><span>            freq (int, optional): Frequency of logging. Defaults to 2.</span><br></span><span><span>        """</span><span></span><br></span><span><span>        </span><span>super</span><span>(</span><span>)</span><span>.</span><span>__init__</span><span>(</span><span>)</span><span></span><br></span><span><span>        self</span><span>.</span><span>trainer </span><span>=</span><span> trainer</span><br></span><span><span>        self</span><span>.</span><span>tokenizer </span><span>=</span><span> tokenizer</span><br></span><span><span>        self</span><span>.</span><span>sample_dataset </span><span>=</span><span> val_dataset</span><span>.</span><span>select</span><span>(</span><span>range</span><span>(</span><span>num_samples</span><span>)</span><span>)</span><span></span><br></span><span><span>        self</span><span>.</span><span>freq </span><span>=</span><span> freq</span><br></span><span><span></span><br></span><span><span>    </span><span>def</span><span> </span><span>on_evaluate</span><span>(</span><span>self</span><span>,</span><span> args</span><span>,</span><span> state</span><span>,</span><span> control</span><span>,</span><span> </span><span>**</span><span>kwargs</span><span>)</span><span>:</span><span></span><br></span><span><span>        </span><span>super</span><span>(</span><span>)</span><span>.</span><span>on_evaluate</span><span>(</span><span>args</span><span>,</span><span> state</span><span>,</span><span> control</span><span>,</span><span> </span><span>**</span><span>kwargs</span><span>)</span><span></span><br></span><span><span>        </span><span># control the frequency of logging by logging the predictions</span><span></span><br></span><span><span>        </span><span># every `freq` epochs</span><span></span><br></span><span><span>        </span><span>if</span><span> state</span><span>.</span><span>epoch </span><span>%</span><span> self</span><span>.</span><span>freq </span><span>==</span><span> </span><span>0</span><span>:</span><span></span><br></span><span><span>            </span><span># generate predictions</span><span></span><br></span><span><span>            predictions </span><span>=</span><span> self</span><span>.</span><span>trainer</span><span>.</span><span>predict</span><span>(</span><span>self</span><span>.</span><span>sample_dataset</span><span>)</span><span></span><br></span><span><span>            </span><span># decode predictions and labels</span><span></span><br></span><span><span>            predictions </span><span>=</span><span> decode_predictions</span><span>(</span><span>self</span><span>.</span><span>tokenizer</span><span>,</span><span> predictions</span><span>)</span><span></span><br></span><span><span>            </span><span># add predictions to a wandb.Table</span><span></span><br></span><span><span>            predictions_df </span><span>=</span><span> pd</span><span>.</span><span>DataFrame</span><span>(</span><span>predictions</span><span>)</span><span></span><br></span><span><span>            predictions_df</span><span>[</span><span>"epoch"</span><span>]</span><span> </span><span>=</span><span> state</span><span>.</span><span>epoch</span><br></span><span><span>            records_table </span><span>=</span><span> self</span><span>.</span><span>_wandb</span><span>.</span><span>Table</span><span>(</span><span>dataframe</span><span>=</span><span>predictions_df</span><span>)</span><span></span><br></span><span><span>            </span><span># log the table to wandb</span><span></span><br></span><span><span>            self</span><span>.</span><span>_wandb</span><span>.</span><span>log</span><span>(</span><span>{</span><span>"sample_predictions"</span><span>:</span><span> records_table</span><span>}</span><span>)</span><span></span><br></span><span><span></span><br></span><span><span></span><br></span><span><span></span><span># First, instantiate the Trainer</span><span></span><br></span><span><span>trainer </span><span>=</span><span> Trainer</span><span>(</span><span></span><br></span><span><span>    model</span><span>=</span><span>model</span><span>,</span><span></span><br></span><span><span>    args</span><span>=</span><span>training_args</span><span>,</span><span></span><br></span><span><span>    train_dataset</span><span>=</span><span>lm_datasets</span><span>[</span><span>"train"</span><span>]</span><span>,</span><span></span><br></span><span><span>    eval_dataset</span><span>=</span><span>lm_datasets</span><span>[</span><span>"validation"</span><span>]</span><span>,</span><span></span><br></span><span><span></span><span>)</span><span></span><br></span><span><span></span><br></span><span><span></span><span># Instantiate the WandbPredictionProgressCallback</span><span></span><br></span><span><span>progress_callback </span><span>=</span><span> WandbPredictionProgressCallback</span><span>(</span><span></span><br></span><span><span>    trainer</span><span>=</span><span>trainer</span><span>,</span><span></span><br></span><span><span>    tokenizer</span><span>=</span><span>tokenizer</span><span>,</span><span></span><br></span><span><span>    val_dataset</span><span>=</span><span>lm_dataset</span><span>[</span><span>"validation"</span><span>]</span><span>,</span><span></span><br></span><span><span>    num_samples</span><span>=</span><span>10</span><span>,</span><span></span><br></span><span><span>    freq</span><span>=</span><span>2</span><span>,</span><span></span><br></span><span><span></span><span>)</span><span></span><br></span><span><span></span><br></span><span><span></span><span># Add the callback to the trainer</span><span></span><br></span><span><span>trainer</span><span>.</span><span>add_callback</span><span>(</span><span>progress_callback</span><span>)</span><br></span>

For a more detailed example please refer to this colab

Additional W&B settings

Further configuration of what is logged with Trainer is possible by setting environment variables. A full list of W&B environment variables can be found here.

Environment VariableUsage
WANDB_PROJECTGive your project a name (huggingface by default)
WANDB_LOG_MODEL
Log the model checkpoint as a W&B Artifact (false by default)
  • false (default): No model checkpointing
  • checkpoint: A checkpoint will be uploaded every args.save_steps (set in the Trainer’s TrainingArguments).
  • end: The final model checkpoint will be uploaded at the end of training.

| | WANDB_WATCH |

Set whether you’d like to log your models gradients, parameters or neither

  • false (default): No gradient or parameter logging
  • gradients: Log histograms of the gradients
  • all: Log histograms of gradients and parameters

| | WANDB_DISABLED | Set to true to disable logging entirely (false by default) | | WANDB_SILENT | Set to true to silence the output printed by wandb (false by default) |

  • Command Line
  • Notebook
<span><span>WANDB_WATCH=all</span><br></span><span><span>WANDB_SILENT=true</span><br></span>

Customize wandb.init

The WandbCallback that Trainer uses will call wandb.init under the hood when Trainer is initialized. You can alternatively set up your runs manually by calling wandb.init before theTrainer is initialized. This gives you full control over your W&B run configuration.

An example of what you might want to pass to init is below. For more details on how to use wandb.init, check out the reference documentation.

<span><span>wandb</span><span>.</span><span>init</span><span>(</span><span></span><br></span><span><span>    project</span><span>=</span><span>"amazon_sentiment_analysis"</span><span>,</span><span></span><br></span><span><span>    name</span><span>=</span><span>"bert-base-high-lr"</span><span>,</span><span></span><br></span><span><span>    tags</span><span>=</span><span>[</span><span>"baseline"</span><span>,</span><span> </span><span>"high-lr"</span><span>]</span><span>,</span><span></span><br></span><span><span>    group</span><span>=</span><span>"bert"</span><span>,</span><span></span><br></span><span><span></span><span>)</span><br></span>

Highlighted Articles

Below are 6 Transformers and W&B related articles you might enjoy

Hyperparameter Optimization for Hugging Face Transformers Hugging Tweets: Train a Model to Generate Tweets Sentence Classification With Hugging Face BERT and WB A Step by Step Guide to Tracking Hugging Face Model Performance Early Stopping in HuggingFace - Examples How to Fine-Tune Hugging Face Transformers on a Custom Dataset

Issues, questions, feature requests

For any issues, questions, or feature requests for the Hugging Face W&B integration, feel free to post in this thread on the Hugging Face forums or open an issue on the Hugging Face Transformers GitHub repo.

Graph View

Backlinks

  • No backlinks found