[BUG]: App crashes with CUDA error in ggml-cuda.cu:1503

[lostmsu]

Description

CUDA error: operation failed due to a previous error during capture
current device: 0, in function ggml_cuda_op_mul_mat at D:\a\LLamaSharp\LLamaSharp\ggml\src\ggml-cuda.cu:1503
cudaGetLastError()
GGML_ASSERT: D:\a\LLamaSharp\LLamaSharp\ggml\src\ggml-cuda.cu:101: !"CUDA error"

From Event Viewer:

c0000409
000000000007f6fe

Reproduction Steps

I'm not sure what causes it, I simply run https://github.com/sangyuxiaowu/LLamaWorker with Meta-Llama-3-8B-Instruct-Q8_0.gguf and after some requests the app crashes with the above error.

Environment & Configuration

• Operating system: Windows 11 23H2
• .NET runtime version: 8.0
• LLamaSharp version: 0.14
• CUDA version (if you are using cuda backend): 12.5.1
• CPU & GPU device: Ryzen 5800 + RTX 3090

Known Workarounds

No response