Adam Optimizer is Not Working in the Fine-Tuning a Pretrained Model Notebook

[arjunaskykok]

This is the notebook:

https://github.com/huggingface/notebooks/blob/main/transformers_doc/en/training.ipynb

In this code cell:

from transformers import TFAutoModelForSequenceClassification
from tensorflow.keras.optimizers import Adam

# Load and compile our model
model = TFAutoModelForSequenceClassification.from_pretrained("bert-base-cased")
# Lower learning rates are often better for fine-tuning transformers
model.compile(optimizer=Adam(3e-5))  # No loss argument!

model.fit(tokenized_data, labels)

We would get this error:

[arjunaskykok]

Passing string works.

model.compile(optimizer="adam")

[arjunaskykok]

This one works as well.

from tf_keras.src.optimizers.legacy.adam import Adam
model.compile(optimizer=Adam(3e-5))

I executed the notebook in Google Colab. The original Adam is an instance of this class:

<keras.src.optimizers.adam.Adam at 0x7d1ec600d790>

Apparently, that class created a problem.