Automatic1111 doesn't start after loading model.

[Neverdusk]

I seem to keep getting this output log after running every step. I don't receive any link or server after running the Colab.

For reference, I inserted a drive folder path into the Model Load step, using a folder that contains multiple models.

After this failed a few times, I tried using the shared link method, but that also produced the same result. Every step previously succeeded, and the repo was confirmed to be up to date in the previous step. This is my first time using this Colab, so I'm not quite sure if I'm doing anything wrong, but any help is appreciated.

[askiiart]

Do you have a token? If not, you'll need that.

[Neverdusk]

I thought the token was only for downloading from Huggingface? I've been trying to use the path_to_trained_model or link_to_trained_model methods. Since I already have my models uploaded to Google Drive.

[askiiart]

You need it every time, sorry. I honestly don't know why, but you do.

[Neverdusk]

I input a token but received the same error. Does the Colab only work with Stable Diffusion 1.4/1.5 and not other models?

[zakpurp]

I think so. Use stable diffusion 1.5 for me it works well

[zakpurp]

Go on hugging face and make sure you accept term to use stable-diffusion-v1. 5

[Neverdusk]

I've already accepted the terms, but I'm trying to use Waifu Diffusion 1.2, since it has an interesting hybrid style.

Is the Colab unable to use the alternate models I already have in Google Drive? Or any custom models?

[askiiart]

Only dreambooth can use different models

[Neverdusk]

Got it. Thanks for the help.

[TheLastBen]

The error is because you're out of RAM (not VRAM), use the 4.7GB CKPT version and it'll work, you can use any model you want. you need the token only when you download the original 1.5 model

[Neverdusk]

Thank you, I'll try with a smaller model then.