mm-eval/TableVQA-Bench
收藏资源简介:
--- dataset_info: features: - name: id dtype: string - name: media list: image - name: messages dtype: string splits: - name: vwtq num_bytes: 138699913 num_examples: 750 - name: vwtq_syn num_bytes: 50271354 num_examples: 250 - name: vtabfact num_bytes: 54083822 num_examples: 250 - name: fintabnetqa num_bytes: 34292847 num_examples: 250 download_size: 277193665 dataset_size: 277347936 configs: - config_name: default data_files: - split: vwtq path: data/vwtq-* - split: vwtq_syn path: data/vwtq_syn-* - split: vtabfact path: data/vtabfact-* - split: fintabnetqa path: data/fintabnetqa-* ---
This is a multimodal dataset that includes images and text messages, designed for tasks such as visual question answering, table fact verification, and financial table question answering. The dataset is divided into four subsets: vwtq (visual question answering), vwtq_syn (synthetic visual question answering), vtabfact (table fact verification), and fintabnetqa (financial table question answering), each containing hundreds of examples, with a total data size of approximately 277MB.




