import thunderkittens error

[klxy0304]

Hey, really awesome repo, I followed the steps in the README to install the pre-existing kernels. After executing python setup.py install, the output was as follows:

Installed /opt/conda/lib/python3.11/site-packages/thunderkittens-0.0.0-py3.11-linux-x86_64.egg
Processing dependencies for thunderkittens==0.0.0
Finished processing dependencies for thunderkittens==0.0.0

However, when I tried to import thunderkittens as tk, I received the following error:

ImportError: /opt/conda/lib/python3.11/site-packages/thunderkittens.cpython-311-x86_64-linux-gnu.so: undefined symbol: _Z17attention_forwardN2at6TensorES0_S0_b

My env is torch==2.5.0, cuda==12.4, GPU: 4090.