search

Zhendong-Wang / Prompt-Diffusion

Official PyTorch implementation of the paper "In-Context Learning Unlocked for Diffusion Models"
Apache License 2.0
364 stars 10 forks source link

How to prepare the whole dataset for training? #3

Open Hongbin98 opened 1 year ago

Hongbin98 commented 1 year ago

Hi! First of all, thank you for your excellent work!

When I try to train a Prompt Diffusion model, I meet the FileNotFoundError. Even if I have run the 'annotate_data.py' to prepare the dataset in default settings, I still get into this trouble. Would you offer me a hand?

Are there any additional requirements for the dataset preparation, or am I mistaken?

Zhendong-Wang commented 1 year ago

Hi there,

Thanks for your interest! Could you provide the error message? I am not sure what files are missing when you were running. For running annotate_data.py, you need to provide the data path, and the seed start and end number for annotation.

Hongbin98 commented 1 year ago

For example, when I try to train a Prompt Diffusion model, I meet the FileNotFoundError: No such file or directory: './clip-filtered-dataset/0409092/2158047904_1_depth.jpg'. And I also find that the annotate_data.py only transforms the images f"{seed}_0_depth.jpg" (only transform '0' but not '1'). Are there any problems with this script?

Zhendong-Wang commented 1 year ago

The annotate_data.py is an example script that we used for data annotation. It shows how to annotate data for {seed}_0. You could easily modify it to annotate {seed}_1 data as well.