Trouble iterating through list of tensors with different ranks in @tf.function method

Question

I'm trying to operate on a list of tensors in a method using a for loop. My goal is to have this loop work as part of a @tf.function. The loop works as epected when running in eager mode, but I get the following error when decorating the method with @tf.function. I am pretty new to Tensorflow. Any help is apprecaited. Thank you!

Code below is written using TensorFlow 2.14

import tensorflow as tf
a = tf.constant([1.,2.])
b = tf.constant([[3.,4.],[5.,6.],[7.,8.]])

tensor_list = [a,b]
multiplier = 2.

@tf.function
def FiniteCheck(list, x):

    isFinite=tf.constant(True)

    for i in tf.range(len(list)):
        isFinite = tf.math.reduce_all(tf.math.is_finite(tf.multiply(list[i], x)))

        if tf.logical_not(isFinite):
            break

    return isFinite

b_Finite =  FiniteCheck(tensor_list, multiplier)

Error being thrown by Tensorflow is here:

File "/Users/xxxx/Desktop/SandBox.py", line 14, in FiniteCheck  *
        isFinite = tf.math.reduce_all(tf.math.is_finite(tf.multiply(list[i], x)))

    TypeError: list indices must be integers or slices, not SymbolicTensor
  File "/var/folders/p1/gftjm5s15mxbxy97573_vn_m0000gn/T/__autograph_generated_fileteduovqt.py", line 49, in tf__FiniteCheck
    ag__.for_stmt(ag__.converted_call(ag__.ld(tf).range, (ag__.converted_call(ag__.ld(len), (ag__.ld(list),), None, fscope),), None, fscope), extra_test, loop_body, get_state_1, set_state_1, ('isFinite', 'break_'), {'iterate_names': 'i'})
  File "/var/folders/p1/gftjm5s15mxbxy97573_vn_m0000gn/T/__autograph_generated_fileteduovqt.py", line 25, in loop_body
    isFinite = ag__.converted_call(ag__.ld(tf).math.reduce_all, (ag__.converted_call(ag__.ld(tf).math.is_finite, (ag__.converted_call(ag__.ld(tf).multiply, (ag__.ld(list)[ag__.ld(i)], ag__.ld(x)), None, fscope),), None, fscope),), None, fscope)
TypeError: list indices must be integers or slices, not SymbolicTensor

During handling of the above exception, another exception occurred:

  File /var/folders/p1/gftjm5s15mxbxy97573_vn_m0000gn/T/__autograph_generated_fileteduovqt.py", line 25, in loop_body
    isFinite = ag__.converted_call(ag__.ld(tf).math.reduce_all, (ag__.converted_call(ag__.ld(tf).math.is_finite, (ag__.converted_call(ag__.ld(tf).multiply, (ag__.ld(list)[ag__.ld(i)], ag__.ld(x)), None, fscope),), None, fscope),), None, fscope)
  File "/var/folders/p1/gftjm5s15mxbxy97573_vn_m0000gn/T/__autograph_generated_fileteduovqt.py", line 49, in tf__FiniteCheck
    ag__.for_stmt(ag__.converted_call(ag__.ld(tf).range, (ag__.converted_call(ag__.ld(len), (ag__.ld(list),), None, fscope),), None, fscope), extra_test, loop_body, get_state_1, set_state_1, ('isFinite', 'break_'), {'iterate_names': 'i'})
  File "/Users/chris/Desktop/SandBox.py", line 21, in <module>
    b_Finite =  FiniteCheck(tensor_list, multiplier)

TypeError: in user code:

    File "/Users/xxxx/Desktop/SandBox.py", line 14, in FiniteCheck  *
        isFinite = tf.math.reduce_all(tf.math.is_finite(tf.multiply(list[i], x))

)

    TypeError: list indices must be integers or slices, not SymbolicTensor

I tried creating a tensor of tensors using tensorflow stacks or ragged tensors and it did not work with different rank tensors. I believe the solution is to get tensor_list to be a tensor of tensors so the index i, which is of type tensor, can be used to access items in the list.