Polyaxon allows to schedule experiments based on hugging-face's transformers library, and supports tracking metrics, outputs, and models.
With Polyaxon you can:
- log hyperparameters for every run
- see learning curves for losses and metrics during training
- see hardware consumption and stdout/stderr output during training
- log images, charts, and other assets
- log git commit information
- log env information
- log model
Polyaxon provides a tracking API to track experiment and report metrics, artifacts, logs, and results to the Polyaxon dashboard.
You can use the tracking API to create a custom tracking experience with Keras.
In order to use Polyaxon tracking with transformers, you need to install Polyaxon library
pip install polyaxon
This is an optional step if you need to perform some manual tracking or to track some information before passing the callback.
from polyaxon import tracking tracking.init(...)
Polyaxon provides a hugging-face callback, you can use this callback with your experiment to report metrics automatically
from polyaxon import tracking from polyaxon.tracking.contrib.hugging_face import PolyaxonCallback # ... tracking.init() #... trainer = Trainer( model=model, args=training_args, train_dataset=train_dataset if training_args.do_train else None, eval_dataset=eval_dataset if training_args.do_eval else None, callbacks=[PolyaxonCallback], # ... )
If you have a run that is already instanciated, you can pass it directly to the callback:
# ... callback = PolyaxonCallback(run=run) trainer = Trainer( model=model, args=training_args, train_dataset=train_dataset if training_args.do_train else None, eval_dataset=eval_dataset if training_args.do_eval else None, callbacks=[callback], # ... )
If you want to have more control and use Polyaxon to log metrics in your custom Keras training loops:
- log metrics
tracking.log_mtrics(metric1=value1, metric2=value2, ...)