Small-tailqwq


Guys, I found the solution, even after the update you can still solve it by the original method. Try deleting the ` map_location=shared.weight_load_location` in `pl_sd = torch.load(checkpoint_file, map_location=shared.weight_load_location)` near line...

Yes, I got the same image as you described the first few times I didn't uninstall, and after uninstalling I finally got a normal image.

me too,in colab

Use `nvcc -V` to check that your cuda is installed correctly and that the version is compatible with torch. If you find that the cuda version using `nvidia-smi` is different...

经过测试，如果你的 png 原图是 8 bit 的话，它的输出结果将会是绿底，如果是 64 bit 的话，就会出现你这样的黑底。解决方法是将图像调整为 32 bit 进行处理。 如果你想批量转换，你可以使用 waifu2x-caffe，选择仅降噪，降噪等级选择 0（确保不对图像内容进行修改），**输出色彩深度** 选 8 即可把图片转换为 32 bit。

双手打是吧，笑死

经过验证，已下载的版权本通过下载界面点击封面显示解析失败，点击标题等其他部位依旧能够正常观看。还请补充更多信息以用作排查。

> I've got the same issue with it failing startup even though I have the appropriate port forwarded. I get this error whether I'm using the port or -net host...