search

FoundationVision / VAR

[NeurIPS 2024 Oral][GPT beats diffusion🔥] [scaling laws in visual generation📈] Official impl. of "Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction". An *ultra-simple, user-friendly yet state-of-the-art* codebase for autoregressive image generation!
MIT License
4.28k stars 314 forks source link

How can I train VAR with my own data? Does the num_class parameter in VAR have any effect? #82

Open zhangqingwu opened 3 months ago

zhangqingwu commented 3 months ago

When building training data and VAR models, there are image-net categories. How should I deal with it when I want to train VAR with my own data?

DanielKahneman commented 2 months ago

I also want to ask this question . I tried to make my dataset more similar to the imagenet dataset, but I still can't train it. I have resized the images to 256*256 and also subdivided them into multiple subfolders like the imagenet train file path. Please tell me what else needs to be adjusted. This is very important to me. Thank you.