一个有趣的事情

目前我是按照版本要求配置的环境python=3.9 torch=2.0.1 cuda=11.3 gpu=2张3090 ubuntu20 我发现当我单卡训练不报错 双卡训练时候，我的batch最大正常训练数量是64。无论我选择amp=true or false 再大batch就报错 RuntimeError: CUDA error: device-side assert triggered CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. Compile with TORCH_USE_CUDA_DSA to enable device-side assertions