search

hutaiHang / Faster-Diffusion

Official implementation of "Faster Diffusion: Rethinking the Role of UNet Encoder in Diffusion Models"
https://arxiv.org/abs/2312.09608
Apache License 2.0
283 stars 19 forks source link

AttributeError: 'UNet2DConditionModel' object has no attribute 'order' #1

Open thanhhung0112 opened 9 months ago

sen-mao commented 9 months ago

No description provided.

Is the version of diffusers you are using 0.17.1 ?

hutaiHang commented 9 months ago

I'm sorry you encountered this problem, our code is too hasty.

Please note the code

image = pipe.call(prompt).images[0]

Not

image = pipe(prompt).images[0]

Because we rewrite the pipeline's call function. If you encounter any other issues, please provide more detailed description.

thanhhung0112 commented 9 months ago

Oh, this is my fault, I commented the line registering the pipeline, but i encounter another issue, TypeError: check_inputs() takes from 5 to 8 positional arguments but 9 were given. I don't know why, I try to check the check_inputs method in the pipeline, it still passes 8 arguments. Can you tell me what's going on?

hutaiHang commented 9 months ago

Oh, this is my fault, I commented the line registering the pipeline, but i encounter another issue, TypeError: check_inputs() takes from 5 to 8 positional arguments but 9 were given. I don't know why, I try to check the check_inputs method in the pipeline, it still passes 8 arguments. Can you tell me what's going on?

Please update the version diffusersto 0.23.0, we have updated the requirement.txt. Thanks for your feedback.

thanhhung0112 commented 9 months ago

Oh, this is my fault, I commented the line registering the pipeline, but i encounter another issue, TypeError: check_inputs() takes from 5 to 8 positional arguments but 9 were given. I don't know why, I try to check the check_inputs method in the pipeline, it still passes 8 arguments. Can you tell me what's going on?

Please update the version diffusersto 0.23.0, we have updated the requirement.txt. Thanks for your feedback.

It works, thank you so much!!!, but i have another question, i try using your method with my personalized SD weight, but the generated image is quite bad in details of eyes. Have you ever seen this problem?

hutaiHang commented 9 months ago

Oh, this is my fault, I commented the line registering the pipeline, but i encounter another issue, TypeError: check_inputs() takes from 5 to 8 positional arguments but 9 were given. I don't know why, I try to check the check_inputs method in the pipeline, it still passes 8 arguments. Can you tell me what's going on?

Please update the version diffusersto 0.23.0, we have updated the requirement.txt. Thanks for your feedback.

It works, thank you so much!!!, but i have another question, i try using your method with my personalized SD weight, but the generated image is quite bad in details of eyes. Have you ever seen this problem?

Perhaps you need to adjust the keytime hyperparameter, similar to issue #7.