five

cua-verse/Aguvis

收藏
Hugging Face2026-04-18 更新2026-04-26 收录
下载链接:
https://hf-mirror.com/datasets/cua-verse/Aguvis
下载链接
链接失效反馈
官方服务:
资源简介:
# cua-verse/Aguvis cua-lite preprocessed version of Aguvis (xlangai/aguvis-stage1 + aguvis-stage2) merged into one repo. Stage-1 contributes grounding:action sub-datasets (OmniAct, RICO, UI-RefExp, GUIEnv, SeeClick, WebUI); stage-2 contributes trajectory data (AITW, Android-Control, CoAT, GUIDE, MiniWoB). Both stages share the unified cua-lite SFT schema; the original stage boundary is preserved in metadata.others. ## Origin - [https://huggingface.co/datasets/xlangai/aguvis-stage1](https://huggingface.co/datasets/xlangai/aguvis-stage1) - [https://huggingface.co/datasets/xlangai/aguvis-stage2](https://huggingface.co/datasets/xlangai/aguvis-stage2) ## Schema Each row has these columns: | column | type | notes | |---|---|---| | `image_ids` | list[string] | content-addressed ids (`<sha256>.<ext>`), used for cross-parquet dedup | | `images` | list[Image] | bytes embedded at HF push time; matches `image_ids` index-for-index | | `messages` | list[struct] | OpenAI-style turns with `role` + structured `content` | | `metadata` | struct | `{platform, task_type, split, others{...}}` | Coordinate values in `messages` are normalized to `[0, 1000]` integers. ## Layout ``` <platform>/<task_type>/<split>/<variant>.parquet ``` - `platform` ∈ {desktop, mobile, web} - `task_type` ∈ {understanding, grounding:action, grounding:point, grounding:bbox, trajectory} - `split` ∈ {train, eval} — `eval` is an in-distribution held-out slice (not used in training) ## Stats | platform | task_type | variant | train | eval | |---|---|---|---:|---:| | desktop | grounding:action | omniact | 5,392 | 99 | | mobile | grounding:action | ricoig16k | 15,774 | 359 | | mobile | grounding:action | ricosca | 171,212 | 2,000 | | mobile | grounding:action | ui_refexp | 15,268 | 356 | | mobile | grounding:action | widget_cap | 99,485 | 1,940 | | mobile | trajectory | aitw | 1,698 | 30 | | mobile | trajectory | android_control | 12,318 | 263 | | mobile | trajectory | coat | 1,306 | 23 | | mobile | trajectory | guide | 595 | 12 | | web | grounding:action | guienv | 325,972 | 2,000 | | web | grounding:action | seeclick | 269,121 | 2,000 | | web | grounding:action | seeclick_mi | 269,119 | 2,000 | | web | grounding:action | webui | 56,302 | 1,087 | | web | trajectory | miniwob | 1,775 | 31 | ## Image store Images are content-addressed by SHA-256 and deduplicated within this repo. The mapping `image_id → bytes` is embedded in every parquet via [`datasets.Image`](https://huggingface.co/docs/datasets/image_load). The `reverse.py` script in [cua-lite/cua-lite](https://github.com/cua-lite/cua-lite/tree/main/scripts/hf_upload) reconstructs a local `image_store/` + reference-only parquets, letting you re-dedup across datasets on your own machine. - Total images: **516,962** - Store size: **209.74 GB** ## Notes Sub-datasets (variants) may have heterogeneous source licenses. See metadata.others.source for provenance. Many Aguvis sub-datasets overlap with standalone cua-lite datasets (Mind2Web, AMEX, etc.); deduplicate before mixing for training. ## License & Citation See original datasets (xlangai/aguvis-stage1 and aguvis-stage2) See https://aguvis-project.github.io/

# cua-verse/Aguvis 本仓库为Aguvis的cua-lite预处理版本,将xlangai/aguvis-stage1与aguvis-stage2两个数据集整合为单一仓库。第一阶段(Stage-1)贡献了**动作锚定(grounding:action)**类子数据集,涵盖OmniAct、RICO、UI-RefExp、GUIEnv、SeeClick、WebUI;第二阶段(Stage-2)提供了轨迹(trajectory)类数据,包括AITW、Android-Control、CoAT、GUIDE、MiniWoB。两个阶段均采用统一的cua-lite监督微调(Supervised Fine-Tuning, SFT)范式,原始的阶段分界信息保留在metadata.others字段中。 ## 数据集来源 - [https://huggingface.co/datasets/xlangai/aguvis-stage1](https://huggingface.co/datasets/xlangai/aguvis-stage1) - [https://huggingface.co/datasets/xlangai/aguvis-stage2](https://huggingface.co/datasets/xlangai/aguvis-stage2) ## 数据结构 每条数据包含以下列: | 列名 | 数据类型 | 说明 | |---|---|---| | `image_ids` | 字符串列表 | 基于内容寻址的ID(格式为`<sha256>.<ext>`),用于跨Parquet文件去重 | | `images` | 图像列表 | 上传至Hugging Face(HF)时嵌入的二进制图像数据,与`image_ids`按索引一一对应 | | `messages` | 结构化列表 | 采用OpenAI格式的对话轮次,包含`role`与结构化的`content`字段 | | `metadata` | 结构体 | 格式为`{platform, task_type, split, others{...}}`的元数据字段 | 对话内容中的坐标值已归一化为`[0, 1000]`区间的整数。 ## 存储布局 <platform>/<task_type>/<split>/<variant>.parquet 存储路径格式为:`<平台>/<任务类型>/<划分>/<变体>.parquet`,其中: - `platform`(平台)可选值为{桌面端(desktop)、移动端(mobile)、网页端(web)} - `task_type`(任务类型)可选值包括{理解(understanding)、锚定点(grounding:point)、边界框锚定(grounding:bbox)、轨迹(trajectory)} - `split`(数据划分)可选值为{训练集(train)、评估集(eval)},其中评估集为分布内的保留子集,不用于模型训练。 ## 数据统计 | 平台 | 任务类型 | 变体 | 训练样本数 | 评估样本数 | |---|---|---|---:|---:| | 桌面端(desktop) | 动作锚定(grounding:action) | omniact | 5,392 | 99 | | 移动端(mobile) | 动作锚定(grounding:action) | ricoig16k | 15,774 | 359 | | 移动端(mobile) | 动作锚定(grounding:action) | ricosca | 171,212 | 2,000 | | 移动端(mobile) | 动作锚定(grounding:action) | ui_refexp | 15,268 | 356 | | 移动端(mobile) | 动作锚定(grounding:action) | widget_cap | 99,485 | 1,940 | | 移动端(mobile) | 轨迹(trajectory) | aitw | 1,698 | 30 | | 移动端(mobile) | 轨迹(trajectory) | android_control | 12,318 | 263 | | 移动端(mobile) | 轨迹(trajectory) | coat | 1,306 | 23 | | 移动端(mobile) | 轨迹(trajectory) | guide | 595 | 12 | | 网页端(web) | 动作锚定(grounding:action) | guienv | 325,972 | 2,000 | | 网页端(web) | 动作锚定(grounding:action) | seeclick | 269,121 | 2,000 | | 网页端(web) | 动作锚定(grounding:action) | seeclick_mi | 269,119 | 2,000 | | 网页端(web) | 动作锚定(grounding:action) | webui | 56,302 | 1,087 | | 网页端(web) | 轨迹(trajectory) | miniwob | 1,775 | 31 | ## 图像存储 本仓库内的图像基于SHA-256(SHA-256)算法进行内容寻址,并实现去重。通过[`datasets.Image`](https://huggingface.co/docs/datasets/image_load)工具,每个Parquet文件中均嵌入了`image_id → 图像二进制数据`的映射关系。[cua-lite/cua-lite](https://github.com/cua-lite/cua-lite/tree/main/scripts/hf_upload)仓库中的`reverse.py`脚本可重建本地`image_store/`目录与仅含引用的Parquet文件,支持用户在本地机器上跨数据集进行二次去重。 - 总图像数量:**516,962** - 存储总大小:**209.74 GB** ## 注意事项 各子数据集(变体)可能采用不同的源许可证,相关来源信息可查看metadata.others.source字段。Aguvis的诸多子数据集与独立的cua-lite数据集(如Mind2Web、AMEX等)存在重叠,在混合用于训练前请务必完成去重操作。 ## 许可证与引用 相关许可证与引用信息请查阅原始数据集xlangai/aguvis-stage1与aguvis-stage2。 详情请访问:https://aguvis-project.github.io/
提供机构:
cua-verse
二维码
社区交流群
二维码
科研交流群
商业服务