运行run_rm.sh报错 RuntimeError: CUDA error: device-side assert triggered

[Candy555]

基于llama13B 运行 run_rm.sh报错如下 数据集用的test.json也不行，单机多卡的环境下报错
../aten/src/ATen/native/cuda/Indexing.cu:1146: indexSelectLargeIndex: block: [420,0,0], thread: [29,0,0] Assertion srcIndex < srcSelectDimSize failed.
../aten/src/ATen/native/cuda/Indexing.cu:1146: indexSelectLargeIndex: block: [420,0,0], thread: [30,0,0] Assertion srcIndex < srcSelectDimSize failed.
../aten/src/ATen/native/cuda/Indexing.cu:1146: indexSelectLargeIndex: block: [420,0,0], thread: [31,0,0] Assertion srcIndex < srcSelectDimSize failed.

RuntimeError: CUDA error: device-side assert triggered
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

[shibing624]

for llama, set pad_token_id = 0, later i will fix it.