I am getting issue while Inferencingmodel

[yashsoulpage]

ValueError: Exception encountered when calling TFViTModel.call().

All operation outputs must be tensors. Operation returned a non-tensor. Non-tensor received: TFBaseModelOutputWithPooling(last_hidden_state=<tf.Tensor 'vit_1/layernorm_1/add_2:0' shape=(None, 197, 768) dtype=float32>, pooler_output=<tf.Tensor 'vit_1/pooler_1/dense_1/Tanh:0' shape=(None, 768) dtype=float32>, hidden_states=None, attentions=None)

Arguments received by TFViTModel.call():
• args=('<KerasTensor shape=(None, 3, 224, 224), dtype=float32, sparse=None, name=keras_tensor_74>',)
• kwargs={'training': 'False'}

with this requirements:
Flask == 1.0.2
Pillow == 7.2.0
loguru == 0.5.3
matplotlib == 3.1.1
numpy == 1.19.5
pandas
scikit-learn
tensorflow == 2.8.0
tqdm == 4.64.0
transformers == 4.18.0
python== 3.10