search

CompVis / latent-diffusion

High-Resolution Image Synthesis with Latent Diffusion Models
MIT License
11.66k stars 1.52k forks source link

No such file or directory: 'configs/first_stage_models/vq-f4/model.yaml' #87

Open canqin001 opened 2 years ago

canqin001 commented 2 years ago

Hi,

When I am trying to train a LDM model as "CUDA_VISIBLE_DEVICES=1 python main.py --base configs/latent-diffusion/lsun_bedrooms-ldm-vq-4.yaml -t --gpus 0", there is an error massage as "No such file or directory: 'configs/first_stage_models/vq-f4/model.yaml'".

Does anyone know how to address it?

Thanks,

GaParmar commented 2 years ago

I am running into to same issues

WonwoongCho commented 2 years ago

I changed the value of ckpt_path to be "models/first_stage_models/vq-f4/model.ckpt". But.. if I run training, a message "Restored from models/first_stage_models/vq-f4/model.ckpt with 0 missing and 55 unexpected keys" is printed out. No idea yet what 55 unexpected keys are.

-- The 55 unexpected keys are from loss.perceptual_loss and loss.discriminator.

Junoh-Kang commented 1 year ago

I also have the same issue here. If there are someone who solved this problem, please let me know.

LSC527 commented 1 year ago

bash scripts/download_first_stages.sh Download ckpt first as described in README.

g-jing commented 1 year ago

I have the same issue about unexpected keys. Do you know how to solve this issue?

TianBian95 commented 1 year ago

I solved this issue by revising the ckpt_path as models/first_stage_models/vq-f4/model.ckpt in configs/latent-diffusion/lsun_bedrooms-ldm-vq-4.yaml.

First, you should make sure you have downloaded the ckpt files by running:

bash scripts/download_first_stages.sh

Then, you could find the ckpt file in models/first_stage_models/vq-f4.

leibohan commented 8 months ago

i am experiencing the same problem. everyone mentioned about how to get .ckpt but we did not fine yaml file corresponding