The effect is poor when using the demo for testing.
when I use cmd: python autoregressive/sample/sample_t2i.py --vq-ckpt ./pretrained_models/vq_ds16_t2i.pt --gpt-ckpt ./pretrained_models/t2i_XL_stage2_512.pt --gpt-model GPT-XL --image-size 512
The generated image is as above. Do you have any ideas?
Hi! Thanks for your interest in our work. I will look into it in a few days.
It works fine for me. Please make sure you use the correct t2i_XL_stage2_512.pt checkpoint of NAR rather than LlamaGen.
I'm sure that the model is correct. The relevant operation logs and model files are shown in the above picture.
And it's downloaded from here: https://huggingface.co/yefly/NAR-XL-t2i-stage2/tree/main,then the name is change d to:t2i_XL_stage2_512.pt