Given that the optimal runtime is quadratic, why not just use nested for-loops?

def find_subarrays(arr, target):
    subarrays = []
    for i in range(len(arr)):
        product = 1
        for j in range(i, len(arr)):
            product *= arr[j]
            if product >= target:
                break
            subarrays.append(arr[i : j + 1])
    return subarrays