No kernel image is available for execution on the device #5723

GitHub Link:
https://github.com/google/jax/issues/5723
主要原因是cuda版本和gpu的driver版本不匹配。导致无法并行计算或者编译。解决方法有两种：

1. 跳过并行编程
   XLA_FLAGS=--xla_gpu_force_compilation_parallelism=1
2. 安装合适的Cuda版本，查询链接：
   https://docs.nvidia.com/deploy/cuda-compatibility/index.html
   简单的来说，
   CUDA Toolkit Linux x86_64 Minimum Required Driver Version Windows Minimum Required Driver Version
   CUDA 12.x >=525.60.13 >=527.41
   CUDA 11.x >= 450.80.02* >=452.39*
   CUDA 10.2 >= 440.33 >=441.22
   CUDA 10.1 >= 418.39 >=418.96
   CUDA 10.0 >= 410.48 >=411.31

因此请检查对不对。当然也不一定，比如是450开头的driver还是用cuda10比较靠谱。