Closed ericosmic closed 3 months ago
@ghostxsl 我使用下面的命令进行多卡训练,也会报错没有找的gpu1
python3 -m paddle.distributed.launch --gpus '0,1' tools/train.py -c configs/rec/PP-OCRv3/ch_PP-OCRv3_rec_distillation.yml -o Global.pretrained_model=ppocr_models/pretrain/ch_PP-OCRv3_rec_train/best_accuracy
那这该怎么解决,需要设置全局设备?
建议可以试一下paddle官方发布的docker
Since you haven\'t replied for more than a year, we have closed this issue/pr. If the problem is not solved or there is a follow-up one, please reopen it at any time and we will continue to follow up. 由于您超过一年未回复,我们将关闭这个issue/pr。 若问题未解决或有后续问题,请随时重新打开,我们会继续跟进。
bug描述 Describe the Bug
使用paddle-gpu2.3.2 只能识别到第一张gpu,其他gpu没有被识别到。 服务器上共有3张显卡
但paddle只能识别到第一张gpu
![image](https://github.com/PaddlePaddle/Paddle/assets/35327931/2d104711-ab11-483c-9d56-ecb0c00f7241)
其他补充信息 Additional Supplementary Information
使用export CUDA_VISIBLE_DEVICE=0,1,2 ,也无法解决 No response