ChenDarYen / Key-Locked-Rank-One-Editing-for-Text-to-Image-Personalization

An Pytorch implementation of the paper Key-Locked Rank One Editing for Text-to-Image Personalization
MIT License
76 stars 7 forks source link

training image folder option not present in the configs/perfusion_inference_sd_v2.yaml #14

Open shrutidhange opened 10 months ago

shrutidhange commented 10 months ago

Hi Team,
Great work! I am trying your code. I was able to run your code and try on some customization. v 1 but with v2 I am facing issues. In the configs/perfusion_inference_sd_v2.yaml I don't see the property to mention the data key which is present in the configs/perfusion_custom.yaml.
I tried to add this below data key in the yaml but it throws error. Please let me know why data key is not present in the perfusion_inference_sd_v2.yaml. Please help me on this.

`data: target: main.DataModuleFromConfig params: batch_size: 4 num_workers: 10 wrap: true train: target: perfusion.dataset.PersonalizedBase params: data_root: '/path/to/your/images/' flip_p: 0.5 # 0. when asymmetric prompt_template:

lightning: callbacks: image_logger: target: main.ImageLogger params: frequency_base: 5 batch_frequency: 25 modelcheckpoint_step: params: every_n_train_steps: 50 trainer: max_steps: 400 log_every_n_steps: 1 accumulate_grad_batches: 4 precision: 16`

huitang0404 commented 7 months ago

it is in data_root