LLaMA-2-13b 多卡推理报错 RuntimeError: CUDA 错误: 触发了设备端断言 请使用 `TORCH_USE_CUDA_DSA` 编译,以启用设备端断言。

作者: huling-2创建于 2024年3月5日更新于 2025年2月20日

../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [32,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [33,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [34,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [35,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [36,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [37,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [38,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [39,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [40,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [41,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [42,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [43,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [44,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [45,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [46,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [47,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` 失败。 ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [0,0,0], thread: [48,0,0] 断言 `-sizes[i] <= index && index < sizes[i] && "index out of

内容来源: LlamaChinese/Llama-Chinese