search

FoundationVision / VAR

[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
4.03k stars 302 forks source link

about the training code #5

Closed WeianMao closed 5 months ago

WeianMao commented 6 months ago

thanks for your great work. I see the training code is empty. when it will be release? Thank you!

keyu-tian commented 6 months ago

Thanks @maowayne123. If you're eager to see our model in action, our inference demo is ready for you at https://github.com/FoundationVision/VAR/blob/main/demo_sample.ipynb. I'm currently polishing up the training codes and they'll be ready in just a few days.

luohao123 commented 5 months ago

Hi, am ask does the current training codes only for conditioned iamge geneation rather than text iamge generation?

Does the tokenizer training code release as well?

keyu-tian commented 5 months ago

Our training codes & scripts are released so im gonna close this. Thank you and hope you enjoy playing with VAR!

@luohao123 this codebase is only for label-to-image generation. VAE training codes will be released at https://github.com/FoundationVision/vae-pro soon.