Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are in
stalled properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices…
链接点开显示No dashboards are active for the current data set.
- TensorFlow:2.9.0
- CUDA:12.0
- cuDNN:无
官方对应版本这个好像已经不更新了
猜测原因:版本不对应
tensorflow GPU测试代码:
import tensorflow as tf
print(tf.config.list_physical_devices('GPU'))
未报错,那就不想重装了,试试挨个包安装一下:
cudnn64_8.dll下载地址
三个文件夹里的文件放到相应的*\NVDIA GPU Computing Toolkit\CUDA\v12.0*中
参考
配置
cublas64_12.dll 改为 cublas64_11.dll
cublasLt64_12.dll 改为 cublasLt64_11.dll
cufft64_10.dll 改为 cufft64_10.dll
cusparse64_11.dll 改为 cusparse64_11.dll
参考
不报错了,但还是No dashboards are active for the current data set.,试试更换原语句.\tensorboard --logdir="logs" --port=6007
- logs的路径含中文:否
- 路径错误,无事件文件:否
- 语句去掉引号:否
- 换浏览器:edge、火狐、谷歌均否
- 换端口号
--host=127.0.0.1 --port=8008
:否
参考
最后看了这篇的三种 启动方式,突发奇想把路径换成了绝对路径,好使了。