Dear author, I'm new to the field of image generation. I'm confused about the evaluation method in your work. From my point of view, you seem to use the evaluation.py to evaluate the quality of the npz file generated by VAR. After carefully reading the README in ./LlamaGen/evaluations/c2i, I still wonder which reference file should I use. Should I use the reference file of ImageNet or LSUN? Many thanks!