To upload your Keras models to the Hub, you can use the push_to_hub_keras function.

!pip install huggingface-hub
!huggingface-cli login
from huggingface_hub.keras_mixin import push_to_hub_keras
push_to_hub_keras(model = model, repo_url = "https://huggingface.co/your-username/your-awesome-model")

To load Keras models from the 🤗Hub, use from_pretrained_keras function.

!pip install huggingface-hub
!huggingface-cli login
from huggingface_hub.keras_mixin import from_pretrained_keras
from_pretrained_keras("your-username/your-awesome-model)

If you'd like to upload 🤗Transformers based Keras checkpoints and let us host your metrics interactively in the repo in with TensorBoard, use PushToHubCallback like follows:

!pip install huggingface-hub
!huggingface-cli login
from transformers.keras_callbacks import PushToHubCallback
from tensorflow.keras.callbacks import TensorBoard
tensorboard_callback = TensorBoard(log_dir = "./logs/tensorboard)
push_to_hub_callback = PushToHubCallback(output_dir="./logs", 
										tokenizer=tokenizer,
										hub_model_id=model_id)

callbacks = [tensorboard_callback, push_to_hub_callback]
model.fit(..., callbacks=callbacks, ...)