Tianwei Zhao

ValueError: Input image size (490*490) doesn't match model (336*336).

5

When I ran the example inference code for model xcomposer2-vl-7b provided in the huggingface page: ``` import torch from transformers import AutoModel, AutoTokenizer torch.set_grad_enabled(False) # init model and tokenizer model...

The download inks in vlmeval/dataset/image_mcq.py for R-Bench are wrong, they download an html page instead of tsv file: ``` # R-Bench 'R-Bench-Dis': 'https://huggingface.co/datasets/lcysyzxdxc/R-Bench/blob/main/R-bench-dis.tsv', 'R-Bench-Ref': 'https://huggingface.co/datasets/lcysyzxdxc/R-Bench/blob/main/R-bench-ref.tsv', ``` The correct link should...