creating a numpy array of mixed data type

I have three dictionaries for example:

first_dictionary = {'Feature1': array([0, 0, 1, 0]),
 'Feature2': array([0, 1, 0, 0]),
 'Feature3': array([1, 0, 0])}

second_dictionary = { 'Feature4': array([0., 1.]),
 'Feature5': array([0.]),
 'Feature6': array([0., 1.])}

third_dictionary = {{'Feature7': array([  0.,   0.,   0., 912.,   0.,
          0.,   0.,   0.,   0.,   0.]),
 'Feature8': array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.])}

Real dictionaries have a lot more keys (about 50 in each). I combine numpy array to produce a single 1D numpy array:

output = []

for dictionary in [first_dictionary,second_dictionary,third_dictionary]:
    for key, value in dictionary.items():
        output += list(value)
    
output_array = np.array(output)

However, when I do this the datatype are all messed where as I want to produce a final numpy array which maintains the datatype of original numpy arrays.

So what I get is:

array([  0.,   0.,   1.,   0.,   0.,   1.,   0.,   0.,   1.,   0.,   0.,
         0.,   1.,   0.,   0.,   1.,   0.,   0.,   0., 912.,   0.,   0.,
         0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,
         0.,   0.,   0.,   0.])

Whereas, as you can you first dictionary only have integers so I want it to look like:

array([  0,    0,    1,    0,    0,    1,    0,    0,   1,     0,    0,
         0.,   1.,   0.,   0.,   1.,   0.,   0.,   0., 912.,   0.,   0.,
         0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,   0.,
         0.,   0.,   0.,   0.])

How can I achieve this? Insights will be appreciated. Side note: All of the numpy arrays in all of dictionaries were created using np.ndarray with type set to None.