TensorFlow 1.0和StyleGAN-InvalidArgumentError：无法分配设备进行操作

Question

我正在尝试在装有NVIDIA GeForce GTX 1650的笔记本电脑上运行Style GAN，并且正在遵循tutorial。我能够生成.tfrecord文件。之后，我尝试运行python train.py，但出现以下错误：

Building TensorFlow graph...

.... some warnings here

tensorflow.python.framework.errors_impl.InvalidArgumentError: Cannot assign a device for operation G_synthesis_2/lod: {{node G_synthesis_2/lod}} was explicitly assigned to /device:GPU:1 but available devices are [ /job:localhost/replica:0/task:0/device:CPU:0, /job:localhost/replica:0/task:0/device:GPU:0 ]. Make sure the device specification refers to a valid device.

.... stack of the errors

During handling of the above exception, another exception occurred:

.... stack of the errors

tensorflow.python.framework.errors_impl.InvalidArgumentError: Cannot assign a device for operation G_synthesis_2/lod: node G_synthesis_2/lod (defined at C:\Users\Acer\anaconda3\envs\tf\lib\site-packages\tensorflow_core\python\framework\ops.py:1748)  was explicitly assigned to /device:GPU:1 but available devices are [ /job:localhost/replica:0/task:0/device:CPU:0, /job:localhost/replica:0/task:0/device:GPU:0 ]. Make sure the device specification refers to a valid device.
         [[G_synthesis_2/lod]]

我使用：

• Windows 10
• NVIDIA GeForce GTX 1650
• Python 3.7.9
• tensorflow-gpu==1.15
• CUDA 10.1
• CUDNN 7.65
• Conda虚拟环境。

我的尝试：

1. 检查CUDA的环境变量
2. 修复DLL错误
3. 检查是否可以识别GPU（是）
4. 将CUDA_VISIBLE_DEVICES更改为0和0,1
5. 卸载tensorflow和protobuf（this answer）

有人可以帮忙吗？