City

Huh, odd. The path is correct in the log from the info you provided so that's not it, and if you refreshed the UI then that shouldn't be the issue...

Closing as HunYian DiT should be feature complete now. The issue for lumia is #56 for those that wish to follow it.

@feffy380 Interesting, seems like you're right. I kind-of just assumed it was some funky conversion logic but I guess that explains it. I've [created a quick script](https://gist.github.com/city96/2db92a207f3f93b1170dbd412688f812) to check for...

@kohya-ss Thank you. That solution is indeed more optimal, I'll close my PR and will follow the one you linked instead. Leaving the issue open until that PR gets merged...

>But I need to admit, I have no idea if it's even possible.. I mean, it should be, if you don't mind the quality being sub-par. I only did a...

@MoonMoon82 Hi, a bit late but I finished training some models for stage_b. Also redid the v1/xl stuff to be slightly better. Can you test it if you have the...

Glad you like it. It was just a quick one-off idea I had, and there's definitely improvements that could be made if I had the time. As for using 2.1,...

Just double checked. Stable diffusion v2.1 and v2.0 both come with the same VAE, which is "ft-mse-840000" - the same one people usually use with SDv1.5. This means it is...

>how could this be improved by the way Well, the neural network part would have to be changed. Currently it's just a bunch of [random conv2D layers that look like...

Same way SD does it. safetensor files just store pairs of keys:values (in this case the values are the network weights). You can just add a prefix to all the...