在 pytorch 中使用 tensorboard 可视化模型、数据和训练

Question

关注this Tutorial 使用 torch.utils.tensorboard 在 tensorboard 中可视化图像时出错

from torch.utils.tensorboard import SummaryWriter
writer = SummaryWriter('runs/fashion_mnist_experiment_1')
writer.add_embedding(features,
                metadata=class_labels,
                label_img=images.unsqueeze(1))
writer.close()

错误：

        AttributeError: module 'tensorflow_core._api.v2.io.gfile' has no attribute 'get_filesystem'

Answer 1

尝试添加

writer = SummaryWriter(path)
img_grid = torchvision.utils.make_grid(torch.FloatTensor(imgs)) #imgs represent a batch of images with shape BxCxHxW
writer_add_image('images', img_grid)

来源：https://pytorch.org/tutorials/intermediate/tensorboard_tutorial.html#writing-to-tensorboard