TypeError: unhashable type: 'numpy.ndarray' in array to set

in this def i get TypeError: unhashable type: 'numpy.ndarray' in line retrieved_indices_set = set(retrieved_indices)

def evaluate_retrieval(query_idx, retrieved_indices, relevant_indices):
    # Convert each list element to a tuple
    # Flatten the two-layer list and convert elements to tuples
    arr = np.array(retrieved_indices) #retrieved_indices => 128 * 128 * 3

# Transpose the array and convert it to a list of tuples
    retrieved_indices = tuple(list(map(tuple, np.vstack(arr.T))))
    print(type(retrieved_indices))

# Create a set from the tuples
    retrieved_indices_set = set(retrieved_indices)
    relevant_retrieved = len(retrieved_indices_set.intersection(relevant_indices_set))
    precision = relevant_retrieved / len(retrieved_indices_set) if len(retrieved_indices_set) > 0 else 0
    return precision

I try this but didn't work
retrieved_indices_tuples = tuple(tuple(tuple(pixel) for pixel in row) for row in retrieved_indices)

I solved this problem in this way:

retrieved_indices_tuples = tuple(
    tuple(
        tuple(
            tuple(pixel) for pixel in slice
        ) for slice in row
    ) for row in retrieved_indices
)
    print(type(retrieved_indices_tuples))  # Should print <class 'tuple'>
    print(type(retrieved_indices_tuples[0]))  # Should print <class 'tuple'>
    print(type(retrieved_indices_tuples[0][0]))  # Should print <class 'tuple'>
    print(type(retrieved_indices_tuples[0][0][0]))  # Should print <class 'tuple'>
    print(type(retrieved_indices_tuples[0][0][0][0])) # Should print <class 'int'>