当前运行环境：
python3.10.8 flash_attn-2.6.1+das.opt2.dtk24043 vllm-0.5.0+das.opt4.dtk24043 torch-2.3.0+das.opt2.dtk24043
torchvision-0.16.0+das.opt1.dtk24043

报错如下：
File "/home/cssc/anaconda3/envs/xinference1/lib/python3.10/site-packages/torch/autograd/function.py", line 598, in apply
return super().apply(*args, **kwargs) # type: ignore[misc]
File "/home/cssc/anaconda3/envs/xinference1/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 621, in forward
out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = _flash_attn_varlen_forward(
File "/home/cssc/anaconda3/envs/xinference1/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 89, in _flash_attn_varlen_forward
out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = flash_attn_cuda.varlen_fwd(
RuntimeError: [address=0.0.0.0:35457, pid=149149] HIP error: invalid device function
HIP kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing AMD_SERIALIZE_KERNEL=3.
Compile with TORCH_USE_HIP_DSA  to enable device-side assertions.