A电脑：
显卡：GTX1060
系统环境/System Environment：win10，cuda10.6，cudnn7.6.5
版本号/Version：Paddle：paddlepaddle-gpu==2.2.2 也尝试过 2.3.0
PaddleOCR： paddleocr==2.5.0.3

B电脑：
显卡：GTX3080
系统环境/System Environment：win10，cuda11.1，cudnn8.1.0
版本号/Version：Paddle：paddlepaddle-gpu==2.3.1.post111 也尝试过 2.1.3.post111
PaddleOCR： paddleocr==2.5.0.3

运行指令/Command Code：python tools/train.py -c .\configs\rec\PP-OCRv3\ch_PP-OCRv3_rec_distillation.yml
完整报错/Complete Error Message：

W0726 19:36:58.004176 9376 device_context.cc:404] Please NOTE: device: 0, GPU Compute Capability: 8.6, Driver API Version: 11.5, Runtime API Version: 11.2
W0726 19:36:58.224562 9376 device_context.cc:422] device: 0, cuDNN Version: 8.1.
[2022/07/26 19:37:03] ppocr INFO: load pretrain successful from ./pretrain_model/ch_PP-OCRv3_rec_train/best_accuracy
[2022/07/26 19:37:03] ppocr INFO: train dataloader has 1576 iters
[2022/07/26 19:37:03] ppocr INFO: valid dataloader has 439 iters
[2022/07/26 19:37:03] ppocr INFO: During the training process, after the 0th iteration, an evaluation is run every 500 iterations
PS G:\driving_permits_distinguishaa\PaddleOCR-release-2.5>

运行训练后大概30秒就停下来了，两台电脑情况一致，而且显存均没有溢出，batch_size设定为1也尝试过，情况依旧。