No file name config.json

[IceIce1ce]

Hi authors,

First of all, congratulations on your paper which was published at ICML this year. When I run the image generation script after finishing training the model. It notifies me we we do not have the config.json file as well as no file name diffusion_pytorch_model.bin. Can you help me to solve this problem.

[OrBerebi]

try setting the "trained_with_lora" parameter to True in the ./eval_scripts/models_catalog.py file.

here is an example that seem to work (at least for me):

"sd_dumb": {
"dataset_size": 100, # how many examples were in the dataset. Use either full or number.
"consistency": True, # whether this model was trained with consistency.
"trained_with_lora": True,
"timestep_nature": 50, # level of noise that was present in the training examples. Number should be between 0 and 1000 for SDXL.
"ckpt_path": "/gpfs0/bgu-br/users/berebio/GitHub/ambient-tweedie/laion_low_level/checkpoint-400/", # add local checkpoint path
"desc": "SDXL dumb model.",
},