jxie/language_table-bootstapir_checkpoint_v2

--- dataset_info: features: - name: tracks dtype: binary - name: query_points dtype: binary - name: visibles dtype: binary - name: movements dtype: float64 - name: text dtype: string - name: video dtype: binary - name: camera_motion dtype: bool splits: - name: train num_bytes: 68126781504 num_examples: 442232 download_size: 60762149532 dataset_size: 68126781504 configs: - config_name: default data_files: - split: train path: data/train-* ---This dataset includes various features such as tracks, query_points, visibles, movements, text, video, and camera_motion. The dataset is split into a training set containing 442232 samples. The download size of the dataset is 60762149532 bytes, and the dataset size is 68126781504 bytes.

搜集汇总

背景与挑战

背景概述

该数据集包含7个特征（tracks、query_points、visibles、movements、text、video、camera_motion），数据类型多样，仅有train分割，共442232个样本，总大小约68.1GB，下载大小约60.8GB，适用于多模态学习任务。