Inference results for some of the new vision and language models. These results were generated using the code from VLMEvalKit, and its main objective is to be able to change the default judge without having to rerun the inference again.
You can recreate either the original results… See the full description on the dataset page:
https://huggingface.co/datasets/jefehern/vlmevalkit_inference.
