TensorFlow调用model.fit时加载libcublasLt.so.12失败求助

曾安装CUDA 12.1但TensorFlow无法检测，降级至11.8版本后TensorFlow可识别GPU，但调用model.fit()时触发以下错误：

Could not load library libcublasLt.so.12. Error: libcublasLt.so.12: cannot open shared object file: No such file or directory

其他基础GPU运算（如print(tf.reduce_sum(tf.random.normal([1000, 1000])))）可正常执行，多次重装CUDA、TensorFlow、cuDNN均无效，已设置LD_LIBRARY_PATH指向CUDA 11.8相关目录。

相关环境及检查结果 #

1. TensorFlow版本及设备检测输出：

TensorFlow version: 2.12.0
PhysicalDevice(name='/physical_device:CPU:0', device_type='CPU')
PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')
CUDA version: 11.8
CUDNN version: 8

2. LD_LIBRARY_PATH配置：

$ export | grep LD_
declare -x LD_LIBRARY_PATH="/home/jprzew/.local/lib/python3.8/site-packages/nvidia/cudnn/:/usr/lib/x86_64-linux-gnu:/usr/local/cuda-11.8/lib64/:/usr/local/cuda-11.8/targets/x86_64-linux/lib/"

3. libcublasLt文件查找结果：

$ sudo find / -name "libcublasLt*"
/usr/local/cuda-11.6/targets/x86_64-linux/lib/libcublasLt.so.11.9.2.110
/usr/local/cuda-11.6/targets/x86_64-linux/lib/libcublasLt.so
/usr/local/cuda-11.6/targets/x86_64-linux/lib/libcublasLt_static.a
/usr/local/cuda-11.6/targets/x86_64-linux/lib/stubs/libcublasLt.so
/usr/local/cuda-11.6/targets/x86_64-linux/lib/libcublasLt.so.11
/usr/local/cuda-11.8/targets/x86_64-linux/lib/libcublasLt.so
/usr/local/cuda-11.8/targets/x86_64-linux/lib/libcublasLt_static.a
/usr/local/cuda-11.8/targets/x86_64-linux/lib/stubs/libcublasLt.so
/usr/local/cuda-11.8/targets/x86_64-linux/lib/libcublasLt.so.11.11.3.6
/usr/local/cuda-11.8/targets/x86_64-linux/lib/libcublasLt.so.11

恳请提供解决线索。

内容的提问来源于stack exchange，提问作者Janusz Przewocki