How to add a nested dictionary as an input to tf.data.Dataset.from_tensor_slices

Question

I am trying to load a dataset using the tf.data.Dataset.from_tensor_slices command.

My input is a list of nested dictionaries in the following format:

a_dict = {  'a' : 'blablabla',
            'b' : {
                    'c': (tf.constant([[0.390, 0.146]])),
                    'd': (tf.constant([0]))                   
                  }
         }

b_dict = {  'a' : 'blablabla',
            'b' : {
                    'c': (tf.constant([[0.453, 0.655], [0.345, 0.784]])),
                    'd': (tf.constant([0, 0]))                   
                  }
         }        
     

Update: Deserializing the input won't work, since the command:
(dataset = tf.data.Dataset.from_tensor_slices(pd.DataFrame.from_dict(pd.json_normalize(train_data)).to_dict(orient="list"))) will give an error

"Shapes of all inputs must match:"

Does anyone have an idea how to load data from a list of nested dictionaries with different structure using tf.dataset?

Thank you in advance!