small_train_dataset = tokenized_datasets["train"].shuffle ... If your dataset is small, you can just convert the whole thing to NumPy arrays and pass it to Keras.
確定! 回上一頁