train_processed_annotations_coco

--- dataset_info: features: - name: input_ids sequence: int32 - name: attention_mask sequence: int8 - name: pixel_values sequence: sequence: sequence: float32 split