prompthero

/

openjourney

like 3.04k

I got this error when trying to generate images. My presumption was that this model does not support fp16, so I commented out a line from my code (not dreambooth):

    pipe = DiffusionPipeline.from_pretrained(
        sd_settings_json["diffusion_settings"]["model_id"],
        custom_pipeline="lpw_stable_diffusion",
        #revision="fp16",
        safety_checker=None,
        torch_dtype=torch.float16,
        use_auth_token=token
    )

We're working in a v2 of the model, hopefully it will be released this week

You can do it with Automatic 1111 CheckpointMerger

Any update on v2 of this model? :D

We're working in a v2 of the model, hopefully it will be released this week

Is it now v2?

Any updates on v2 ?

We're working on it, we have some .ckpts but we're still fine-tuning it

would love to know if there's any update on this?

@prompthero :

I see V2 is already available... but why only on .ckpt format?

Both diffuser and some Dreambooth scripts work only with a the vae; unet, text_encoder, etc. bin and json files.

Until you convert the .ckpts to this format, they will be useless for many people...

sorry pretty new to this, I try to load the .ckpt file into the SD models but seems like it refused to load, what would be the correct step?

I am trying to fine tune this using Dreambooth and I'm still getting this error for this model as well as V2. Any idea how I can fix this?

You can do it with Automatic 1111 CheckpointMerger

Should I merge the checkpoints of this with another model like Stable Diffusion v1.5?