Decouple the kernels

[voltjia]

pytest output:

============================= test session starts ==============================
platform linux -- Python 3.10.12, pytest-8.3.3, pluggy-1.5.0
rootdir: /home/huangjiacheng/ntops
configfile: pyproject.toml
plugins: cov-6.0.0
collected 62 items

tests/test_abs.py ........                                               [ 12%]
tests/test_add.py ........                                               [ 25%]
tests/test_addmm.py ..                                                   [ 29%]
tests/test_bmm.py ..                                                     [ 32%]
tests/test_div.py ........                                               [ 45%]
tests/test_exp.py ........                                               [ 58%]
tests/test_gelu.py ........                                              [ 70%]
tests/test_mm.py ..                                                      [ 74%]
tests/test_mul.py ........                                               [ 87%]
tests/test_rsqrt.py ........                                             [100%]

======================== 62 passed in 167.55s (0:02:47) ========================