Add real-world and simulation public test sets in visual LeRobot v3 format

README.md

@@ -1,5 +1,5 @@
 ---
-pretty_name: Value Model Benchmark Training Data
+pretty_name: RoboValue Training and Evaluation Data
 language:
 - en
 - zh
@@ -9,16 +9,28 @@ tags:
 - robovalue
 - value-model-benchmark
 - training-data
+- evaluation-data
+- diagnostic-test-set
 - simulation
 - real-world
 - full-shot
 - one-shot
+configs:
+- config_name: real-world-test-30
+  default: true
+  data_files:
+  - split: test
+    path: evaluation/real-world/test-30/episode_catalog.parquet
+- config_name: simulation-test-30
+  data_files:
+  - split: test
+    path: evaluation/simulation/test-30/episode_catalog.parquet
 ---
 
 <div align="center">
   <img src="https://huggingface.co/datasets/wzlllong/value-model-benchmark-training-data/resolve/main/assets/robovalue-logo.png" alt="RoboValue" width="420" />
-  <h1>Training Data for Robotic Value Models</h1>
-  <p><strong>Simulation &amp; Real-World · Full-Shot &amp; One-Shot</strong></p>
+  <h1>RoboValue Training and Evaluation Data</h1>
+  <p><strong>Training Demonstrations · Small-Scale Real-World &amp; Simulation Public Test Sets</strong></p>
   <p>
     <a href="https://arxiv.org/abs/2610.26663"><img src="https://img.shields.io/badge/arXiv-Paper-red?logo=arxiv" alt="RoboValue paper on arXiv" /></a>
     <a href="https://github.com/RoboValue-Benchmark/RoboValue"><img src="https://img.shields.io/badge/GitHub-RoboValue-24292F?logo=github" alt="RoboValue GitHub" /></a>
@@ -30,10 +42,10 @@ tags:
 
 RoboValue provides expert demonstrations for **robotic value models**, covering **15 simulation tasks** and **20 real-world tasks**. Both domains offer Full-Shot and One-Shot subsets in the shared **LeRobot Dataset v3.0** format.
 
-All four training subsets are publicly available. [The paper](https://arxiv.org/abs/2610.26663) and current leaderboard report **Zero-Shot and One-Shot evaluations**.
+All four training subsets are publicly available. Two **small-scale public diagnostic test sets** are organized separately under `evaluation/`, with **30 real-world trajectories and 30 simulation trajectories**. Each domain has four tasks and 90 synchronized camera videos. Together the test subsets contain **60 trajectories, 44,647 time steps, and 180 videos**. [The paper](https://arxiv.org/abs/2610.26663) and current leaderboard report **Zero-Shot and One-Shot evaluations**.
 
 <p align="center">
-  <a href="#overview">Overview</a> · <a href="#simulation">Simulation</a> · <a href="#real-world">Real-World</a> · <a href="#repository-layout">Repository</a> · <a href="#download-and-reading">Download &amp; Read</a>
+  <a href="#overview">Overview</a> · <a href="#simulation">Simulation</a> · <a href="#real-world">Real-World Training</a> · <a href="#public-evaluation">Public Test Sets</a> · <a href="#repository-layout">Repository</a> · <a href="#download-and-reading">Download &amp; Read</a>
 </p>
 
 ## 📊 1. Dataset Overview <a name="overview"></a>
@@ -48,11 +60,36 @@ All four training subsets are publicly available. [The paper](https://arxiv.org/
 | ▦ **Full-Shot** | 100 per task | Training or fine-tuning on the complete expert collection |
 | ◉ **One-Shot** | 1 per task | Reference conditioning or task-specific adaptation |
 
-Across all four subsets, the release contains **3,535 episodes, 2,348,375 synchronized time steps, and 10,605 camera videos**. The two One-Shot subsets account for **35 episodes, 24,442 time steps, and 105 camera videos**.
+Across all four training subsets, the release contains **3,535 episodes, 2,348,375 synchronized time steps, and 10,605 camera videos**. The two One-Shot subsets account for **35 episodes, 24,442 time steps, and 105 camera videos**.
 
-> **HF preview:** The automatic preview counts video files as rows, including one showcase clip. Use the episode and time-step totals here and the [LeRobot loader](#download-and-reading) for training.
 
-## 🟦 2. Simulation <a name="simulation"></a>
+> **HF preview:** The explicit `real-world-test-30/test` and `simulation-test-30/test` configurations each contain **one row per test trajectory (30 rows)**. Select a configuration to browse its task, trajectory type, applicable metrics, and three camera video paths. Real-world is the default. These catalogs are browsing indices; native per-frame model input is read with LeRobot. Auxiliary JSON reports and navigation documents are excluded from the preview.
+
+### Public Test Sets <a name="public-evaluation"></a>
+
+Training demonstrations and public test trajectories have separate directories and purposes:
+
+| Collection | Location | Scale | Purpose |
+| --- | --- | --- | --- |
+| Simulation / real-world training | `data/<domain>/{full-shot,one-shot}` | Four training subsets listed above | Expert demonstrations for training or reference conditioning |
+| [Real-world public test set](evaluation/real-world/test-30/README.md) | `evaluation/real-world/test-30` | **4 tasks / 30 trajectories / 27,498 time steps / 90 videos** | Small-scale In-Domain diagnostic evaluation |
+| [Simulation public test set](evaluation/simulation/test-30/README.md) | `evaluation/simulation/test-30` | **4 tasks / 30 trajectories / 17,149 time steps / 90 videos** | Small-scale In-Domain diagnostic evaluation |
+
+The real-world subset contains **12 desk**, **12 flower-and-diffuser**, **3 drum**, and **3 button** trajectories. The simulation subset contains **12 pen-holder**, **12 mug-hanging**, **3 press-by-number**, and **3 block-swapping** trajectories. Both supply images and English instructions, with evaluation annotations stored separately. Each subset has its own local task and episode indices, independent of training and of the other test subset. They provide partial diagnostic coverage rather than a complete leaderboard test release.
+
+| Coverage | Real-world | Simulation |
+| --- | --- | --- |
+| TGA-CT, TGA-CF, VOC, Cycle-VOC, VS | Each uses 4 ED episodes | Each uses 4 ED episodes |
+| SA / SIA | 24 / 12 episodes | 24 / 12 episodes |
+| TRR | 12 episodes / 4 three-role groups | 12 episodes / 4 three-role groups |
+| CSVC | 12 episodes / 10 solution groups | 12 episodes / 6 order groups |
+| FPL | 8 ER/EC episodes | Original FPL fields retained; no formal FPL cohort |
+| VOC-MEM | 6 complete counting histories | 6 complete memory histories; no annotated counting-subtask intervals |
+
+See the [real-world metric coverage and trajectory index](evaluation/real-world/test-30/README.md#metric-coverage) and [simulation metric coverage and trajectory index](evaluation/simulation/test-30/README.md#metric-coverage). Simulation ED/MS terminal labels and semantic alignment are user-attested; TRR/MEM terminal-label evidence retains the source status of pending independent verification. The simulation README documents these annotation evidence states.
+
+
+## 🟦 2. Simulation Training <a name="simulation"></a>
 
 ### 2.1 Overview
 
@@ -166,7 +203,7 @@ Loaded images use **CHW (channels, height, width)** order.
 
 **Subtask annotations:** [58 paper-defined segments](https://huggingface.co/datasets/wzlllong/value-model-benchmark-training-data/blob/main/data/simulation/one-shot/annotations/subtasks/README.md) label all **8,080 frames**, with one subtask per frame. These annotations accompany the training demonstrations and are separate from benchmark test-set labels. [`dataset_labeler`-compatible JSON](https://huggingface.co/datasets/wzlllong/value-model-benchmark-training-data/tree/7fd21cb2519b4d59cbe9fe8b23cfdf477e3ad145/data/simulation/one-shot/annotations/subtasks/dataset_labeler) uses the same subtasks and boundaries as the frame labels.
 
-## 🟧 3. Real-World <a name="real-world"></a>
+## 🟧 3. Real-World Training <a name="real-world"></a>
 
 ### 3.1 Overview
 
@@ -316,12 +353,23 @@ data/
 └── real-world/
     ├── full-shot/                2,000 episodes · 20 tasks
     └── one-shot/                 20 episodes · 1 per task
+evaluation/
+├── simulation/test-30/           30 trajectories · 4 tasks
+│   └── meta/, data/, videos/, annotations/, README files and catalog
+└── real-world/test-30/            30 trajectories · 4 tasks
+    ├── meta/                     Native LeRobot v3 metadata
+    ├── data/                     Frame references
+    ├── videos/                   Three synchronized camera streams
+    ├── annotations/              Separate evaluation labels
+    ├── episode_catalog.parquet   One browsing row per trajectory
+    ├── README.md / README_zh.md   Task and metric documentation
+    └── checksums.sha256           File integrity manifest
 ```
 
-Each subset directory is an independent **LeRobot** dataset root (`root`), containing per-step Parquet data, three original camera streams, and metadata.
+Each released training or public test subset directory is an independent **LeRobot** dataset root (`root`). Both test subsets have frame references, RGB videos, and English instructions, with evaluation labels alongside them; they have no robot state/action vectors. The primary view is the head camera for real-world and the front camera for simulation.
 
 <details>
-<summary><strong>Subset files and metadata</strong></summary>
+<summary><strong>Training subset files and metadata</strong></summary>
 
 | File or directory | Purpose |
 | --- | --- |
@@ -352,10 +400,12 @@ Select one `prefix` to download only that subset. Sizes use decimal GB.
 | 🟦 **Simulation One-Shot** | `data/simulation/one-shot` | 15 / 8,080 | ≈0.50 GB |
 | 🟧 **Real-World Full-Shot** | `data/real-world/full-shot` | 2,000 / 1,534,321 | ≈23.13 GB |
 | 🟧 **Real-World One-Shot** | `data/real-world/one-shot` | 20 / 16,362 | ≈0.22 GB |
+| **Real-World Public Test** | `evaluation/real-world/test-30` | 30 / 27,498 | ≈0.38 GB |
+| **Simulation Public Test** | `evaluation/simulation/test-30` | 30 / 17,149 | ≈0.595 GB |
 
 ### 5.2 Download and read
 
-This example loads the smaller **Simulation One-Shot** subset and returns states, actions, and all three camera frames. To switch subsets, change `prefix` to the corresponding path in the table. Public data can be downloaded anonymously, and this example automatically reuses an existing local Hugging Face login. If anonymous downloads are rate-limited, run `hf auth login` and retry.
+This example loads the smaller **Simulation One-Shot** subset and returns states, actions, and all three camera frames. To switch training subsets, change `prefix` to the corresponding training path in the table. For the public test set, use the separate image-and-instruction example below. Public data can be downloaded anonymously, and this example automatically reuses an existing local Hugging Face login. If anonymous downloads are rate-limited, run `hf auth login` and retry.
 
 The examples were tested with **Python 3.12**, **LeRobot 0.4.2**, and **PyAV 15.1.0**:
 
@@ -473,4 +523,35 @@ Each subset contains `checksums.sha256`. Run this command from the subset direct
 shasum -a 256 -c checksums.sha256
 ```
 
+### 5.5 Download and read a public test subset
+
+Choose `evaluation/real-world/test-30` or `evaluation/simulation/test-30` to download one public test subset. The example is pinned to `578593ef28ee7a2d861ebf38162c928291b0d4a9`, which contains both test domains; downloading and loading use the same revision. The original training-data revision shown above predates this test release.
+
+```python
+from pathlib import Path
+from huggingface_hub import snapshot_download
+from lerobot.datasets.lerobot_dataset import LeRobotDataset
+
+repo_id = "wzlllong/value-model-benchmark-training-data"
+prefix = "evaluation/real-world/test-30"
+revision = "578593ef28ee7a2d861ebf38162c928291b0d4a9"
+snapshot = snapshot_download(
+    repo_id=repo_id, repo_type="dataset", revision=revision,
+    allow_patterns=[f"{prefix}/**"], local_dir="./robovalue-data",
+)
+root = Path(snapshot) / prefix
+dataset = LeRobotDataset(
+    repo_id=repo_id, root=root, revision=revision,
+    video_backend="pyav", tolerance_s=1e-4,
+)
+sample = dataset[0]
+value_input = {
+    "images": {camera: sample[camera] for camera in dataset.meta.camera_keys},
+    "instruction": sample["task"],
+}
+print(dataset.num_episodes, dataset.num_frames)  # real-world: 30 27498; simulation: 30 17149
+```
+
+The native metadata retains `train: 0:30` as its LeRobot container split. Both public test subsets are intended for **testing**, and the HF browsing index uses the semantic `test` split. Load evaluation annotations explicitly from `root / "annotations"`; do not pass labels, success flags, trajectory types, metric memberships, or frame references to the model. History-based models can use `delta_timestamps` within each episode. See the [real-world README](evaluation/real-world/test-30/README.md) or [simulation README](evaluation/simulation/test-30/README.md) for annotation joins, metric cohorts, and integrity checks. The catalogs share 12 columns; `video_primary` refers to head/front respectively, and the native camera keys preserve those view meanings.
+
 **Publication authorization:** All data in this dataset were authorized before public release.

README_zh.md

@@ -1,7 +1,7 @@
 <div align="center">
   <img src="https://huggingface.co/datasets/wzlllong/value-model-benchmark-training-data/resolve/main/assets/robovalue-logo.png" alt="RoboValue" width="420" />
-  <h1>机器人价值模型训练数据集</h1>
-  <p><strong>Simulation &amp; Real-World · Full-Shot &amp; One-Shot</strong></p>
+  <h1>RoboValue 训练与公开测试数据</h1>
+  <p><strong>训练示例 · 小规模真机与仿真公开测试集</strong></p>
   <p>
     <a href="https://arxiv.org/abs/2610.26663"><img src="https://img.shields.io/badge/arXiv-Paper-red?logo=arxiv" alt="RoboValue paper on arXiv" /></a>
     <a href="https://github.com/RoboValue-Benchmark/RoboValue"><img src="https://img.shields.io/badge/GitHub-RoboValue-24292F?logo=github" alt="RoboValue GitHub" /></a>
@@ -13,10 +13,10 @@
 
 RoboValue 提供面向**机器人价值模型**的专家示例，覆盖 **15 个仿真任务**和 **20 个真实世界任务**。两个领域均提供 Full-Shot 与 One-Shot，采用统一的 **LeRobot Dataset v3.0** 格式。
 
-本仓库的四个训练子集均已完整公开。[论文](https://arxiv.org/abs/2610.26663)与当前排行榜只报告 **Zero-Shot 和 One-Shot 评测结果**。
+本仓库的四个训练子集均已完整公开。另在独立的 `evaluation/` 目录组织两个**小规模公开诊断测试集**：**真机30条、仿真30条轨迹**，每个领域各含4个任务、90个同步相机视频。两个测试子集合计 **60条轨迹、44,647个时间步和180个视频**。[论文](https://arxiv.org/abs/2610.26663)与当前排行榜只报告 **Zero-Shot 和 One-Shot 评测结果**。
 
 <p align="center">
-  <a href="#overview">数据总览</a> · <a href="#simulation">仿真世界</a> · <a href="#real-world">真实世界</a> · <a href="#repository-layout">仓库目录</a> · <a href="#download-and-reading">下载与读取</a>
+  <a href="#overview">数据总览</a> · <a href="#simulation">仿真世界</a> · <a href="#real-world">真实世界训练数据</a> · <a href="#public-evaluation">公开测试集</a> · <a href="#repository-layout">仓库目录</a> · <a href="#download-and-reading">下载与读取</a>
 </p>
 
 ## 📊 1. 数据总览 <a name="overview"></a>
@@ -31,11 +31,36 @@ RoboValue 提供面向**机器人价值模型**的专家示例，覆盖 **15 个
 | ▦ **Full-Shot** | 每任务 100 条 | 使用完整专家示例进行训练或微调 |
 | ◉ **One-Shot** | 每任务 1 条 | 以单条任务示例进行参考条件输入或任务适配 |
 
-四个子集合计 **3,535 条轨迹、2,348,375 个同步时间步、10,605 个相机视频**，其中 One-Shot 合计 **35 条轨迹、24,442 个时间步、105 个相机视频**。
+四个训练子集合计 **3,535 条轨迹、2,348,375 个同步时间步、10,605 个相机视频**，其中 One-Shot 合计 **35 条轨迹、24,442 个时间步、105 个相机视频**。
 
-> **HF 预览说明：**自动预览按视频文件计数，其中包含 1 个展示视频。训练规模请查看这里的轨迹数与时间步数，并按[LeRobot 读取示例](#download-and-reading)加载数据。
 
-## 🟦 2. 仿真世界 <a name="simulation"></a>
+> **HF 预览说明：**显式配置的 `real-world-test-30/test` 和 `simulation-test-30/test` 均按测试轨迹计数，**每条轨迹一行、每个配置30行**。切换配置可查看任务、轨迹类型、适用指标和三机位视频路径，默认显示真机。这些目录用于浏览；实际逐帧模型输入仍通过 LeRobot 读取。辅助JSON报告与导航说明不进入预览。
+
+### 公开测试集 <a name="public-evaluation"></a>
+
+训练示例与公开测试轨迹分别存放，用途明确区分：
+
+| 数据集合 | 目录 | 规模 | 用途 |
+| --- | --- | --- | --- |
+| 仿真／真机训练数据 | `data/<domain>/{full-shot,one-shot}` | 上述四个训练子集 | 用于训练或示例参考输入的专家示范 |
+| [真机公开测试集](evaluation/real-world/test-30/README_zh.md) | `evaluation/real-world/test-30` | **4任务／30轨迹／27,498时间步／90视频** | 小规模 In-Domain 诊断评测 |
+| [仿真公开测试集](evaluation/simulation/test-30/README_zh.md) | `evaluation/simulation/test-30` | **4任务／30轨迹／17,149时间步／90视频** | 小规模 In-Domain 诊断评测 |
+
+真机子集包含**书桌12条、花艺与香薰12条、敲鼓3条和按钮3条**；仿真子集包含**笔筒12条、悬挂马克杯12条、依数字按按钮3条和交换积木3条**。两域均提供图像与英文指令，评测标注独立存放。任务和轨迹编号仅在各自子集内有效，与训练及另一个测试子集的编号独立。两个子集提供部分诊断覆盖，规模与覆盖范围小于完整排行榜测试集。
+
+| 指标覆盖 | 真机 | 仿真 |
+| --- | --- | --- |
+| TGA-CT、TGA-CF、VOC、Cycle-VOC、VS | 各4条 ED | 各4条 ED |
+| SA／SIA | 24／12条 | 24／12条 |
+| TRR | 12条／4个三角色组 | 12条／4个三角色组 |
+| CSVC | 12条／10个方案组 | 12条／6个顺序组 |
+| FPL | 8条 ER/EC | 原FPL字段保留，未建立正式FPL集合 |
+| VOC-MEM | 6条完整计数历史 | 6条完整记忆历史，未提供计数子任务区间 |
+
+请查看[真机指标覆盖与逐轨迹索引](evaluation/real-world/test-30/README_zh.md#metric-coverage)和[仿真指标覆盖与逐轨迹索引](evaluation/simulation/test-30/README_zh.md#metric-coverage)。仿真 ED/MS 的终局标签和语义对齐来自用户确认；TRR/MEM 的终局标签保留源标注中“待独立验证”的证据状态，详情见仿真说明。
+
+
+## 🟦 2. 仿真训练数据 <a name="simulation"></a>
 
 ### 2.1 总览
 
@@ -149,7 +174,7 @@ Full-Shot 与 One-Shot 中，每条 **episode** 都是一条完整的任务示
 
 **子任务标注：**[58 个论文子任务区间](https://huggingface.co/datasets/wzlllong/value-model-benchmark-training-data/blob/main/data/simulation/one-shot/annotations/subtasks/README.md)覆盖全部 **8,080 帧**，每帧对应一个子任务。标注用于这些训练示例，与评测集标签分开。 另提供 [`dataset_labeler` 兼容 JSON](https://huggingface.co/datasets/wzlllong/value-model-benchmark-training-data/tree/7fd21cb2519b4d59cbe9fe8b23cfdf477e3ad145/data/simulation/one-shot/annotations/subtasks/dataset_labeler)，与逐帧标签使用相同的子任务和边界。
 
-## 🟧 3. 真实世界 <a name="real-world"></a>
+## 🟧 3. 真实世界训练数据 <a name="real-world"></a>
 
 ### 3.1 总览
 
@@ -299,12 +324,23 @@ data/
 └── real-world/
     ├── full-shot/                2,000 条 · 20 个任务
     └── one-shot/                 20 条 · 每任务 1 条
+evaluation/
+├── simulation/test-30/           30条轨迹 · 4个任务
+│   └── meta/、data/、videos/、annotations/、说明与目录
+└── real-world/test-30/            30条轨迹 · 4个任务
+    ├── meta/                     原生 LeRobot v3 元数据
+    ├── data/                     逐帧引用
+    ├── videos/                   三路同步相机视频
+    ├── annotations/              独立评测标注
+    ├── episode_catalog.parquet   每条轨迹一行的浏览索引
+    ├── README.md / README_zh.md   任务与指标说明
+    └── checksums.sha256           文件完整性清单
 ```
 
-每个子集目录都是独立的 **LeRobot** 数据根目录（`root`），内部包含逐时间步 Parquet、三路原始相机视频及元数据。
+已发布的训练及公开测试子集目录均为独立的 **LeRobot** 数据根目录（`root`）。两个测试子集保留逐帧引用、RGB视频和英文指令，评测标注另行存放，不含机器人状态／动作向量。真机主视角是头部，仿真主视角是正面。
 
 <details>
-<summary><strong>子集文件结构与元数据</strong></summary>
+<summary><strong>训练子集文件结构与元数据</strong></summary>
 
 | 文件或目录 | 用途 |
 | --- | --- |
@@ -335,10 +371,12 @@ data/
 | 🟦 **仿真 One-Shot** | `data/simulation/one-shot` | 15 / 8,080 | ≈0.50 GB |
 | 🟧 **真机 Full-Shot** | `data/real-world/full-shot` | 2,000 / 1,534,321 | ≈23.13 GB |
 | 🟧 **真机 One-Shot** | `data/real-world/one-shot` | 20 / 16,362 | ≈0.22 GB |
+| **真机公开测试集** | `evaluation/real-world/test-30` | 30 / 27,498 | ≈0.38 GB |
+| **仿真公开测试集** | `evaluation/simulation/test-30` | 30 / 17,149 | ≈0.595 GB |
 
 ### 5.2 下载并读取
 
-以下示例读取较小的 **仿真 One-Shot**，返回状态、动作及三路相机帧。切换子集时，将 `prefix` 改为上表中的对应路径。公开数据可匿名下载，示例会自动复用本机已有的 Hugging Face 登录。若匿名下载遇到限流，可先运行 `hf auth login` 后重试。
+以下示例读取较小的 **仿真 One-Shot**，返回状态、动作及三路相机帧。切换训练子集时，将 `prefix` 改为上表中的对应训练路径。公开测试集请使用下方单独的图像和指令读取示例。公开数据可匿名下载，示例会自动复用本机已有的 Hugging Face 登录。若匿名下载遇到限流，可先运行 `hf auth login` 后重试。
 
 以下示例已在 **Python 3.12**、**LeRobot 0.4.2** 与 **PyAV 15.1.0** 环境中验证：
 
@@ -456,4 +494,35 @@ batch 包含状态、动作、任务指令与三路相机观测。模型的输
 shasum -a 256 -c checksums.sha256
 ```
 
+### 5.5 下载并读取公开测试子集
+
+选择 `evaluation/real-world/test-30` 或 `evaluation/simulation/test-30`，仅下载其中一个公开测试子集。下方示例固定到包含两个测试集的版本 `578593ef28ee7a2d861ebf38162c928291b0d4a9`，下载和加载使用同一提交。上方训练示例的旧版本早于本测试集发布。
+
+```python
+from pathlib import Path
+from huggingface_hub import snapshot_download
+from lerobot.datasets.lerobot_dataset import LeRobotDataset
+
+repo_id = "wzlllong/value-model-benchmark-training-data"
+prefix = "evaluation/real-world/test-30"
+revision = "578593ef28ee7a2d861ebf38162c928291b0d4a9"
+snapshot = snapshot_download(
+    repo_id=repo_id, repo_type="dataset", revision=revision,
+    allow_patterns=[f"{prefix}/**"], local_dir="./robovalue-data",
+)
+root = Path(snapshot) / prefix
+dataset = LeRobotDataset(
+    repo_id=repo_id, root=root, revision=revision,
+    video_backend="pyav", tolerance_s=1e-4,
+)
+sample = dataset[0]
+value_input = {
+    "images": {camera: sample[camera] for camera in dataset.meta.camera_keys},
+    "instruction": sample["task"],
+}
+print(dataset.num_episodes, dataset.num_frames)  # 真机：30 27498；仿真：30 17149
+```
+
+原生元数据保留 `train: 0:30` 作为 LeRobot 容器划分，两个公开子集的用途均为**测试**，HF 浏览索引使用语义上的 `test` 划分。请显式读取 `root / "annotations"` 中的标注；成功标记、轨迹类型、指标归属和逐帧引用不作为模型观测输入。依赖历史的模型可通过 `delta_timestamps` 在每条轨迹内读取时序窗口。标注关联、指标归属与完整性检查见[真机说明](evaluation/real-world/test-30/README_zh.md)或[仿真说明](evaluation/simulation/test-30/README_zh.md)。两域轨迹目录采用相同12列，`video_primary` 分别指向头部／正面视频；原生相机键保留各自的机位含义。
+
 **发布授权：**本数据集全部数据均经授权后公开发布。

evaluation/real-world/test-30/README.md

@@ -0,0 +1,183 @@
+# RoboValue Small-Scale Real-World Public Test Set
+
+[简体中文](README_zh.md) · [Repository overview](../../../README.md) · [Browse all 30 trajectories](TRAJECTORIES.md) · [Parquet catalog](episode_catalog.parquet)
+
+This **in-domain diagnostic test subset** contains **30 complete real-world trajectories across four tasks**, including standard demonstrations, alternative successful solutions, errors, recoveries, and counting histories. It provides visual inputs, English instructions, and separate evaluation annotations for **11 metric entry points**. Its purpose is to make metric behavior inspectable and support small-scale evaluation; its task and solution coverage is smaller than the complete RoboValue leaderboard test set.
+
+The subset is stored under `evaluation/real-world/test-30/`, separately from the repository's training demonstrations in `data/`. The four tasks belong to existing real-world task categories; evaluation coverage is defined by the metric cohorts below. Native metadata retains `train: 0:30` for LeRobot container compatibility; the release's intended use and HF trajectory-catalog split are **test**.
+
+## Tasks and trajectories
+
+| Local task index | Task | Episodes | Local episode indices |
+|---:|---|---:|---|
+| 0 | Organize Desktop Workstation | 12 | 0–11 |
+| 1 | Arrange Flowers and Reed Diffuser | 12 | 12–23 |
+| 2 | Strike Drum Five Times | 3 | 24–26 |
+| 3 | Press Button Five Times | 3 | 27–29 |
+
+There are **27,498 synchronized time steps at 30 FPS** and **90 HEVC videos**: one head-camera, one left-wrist, and one right-wrist video per episode. Time steps are counted once, rather than once per camera. Head videos are **640 × 400**, and wrist videos are **640 × 528**. The visual-input source package is approximately **382 MB (0.38 GB, decimal)**; the release adds lightweight documentation and catalogs. The subset contains **22 successful and 8 unsuccessful episodes**. Complete trajectories and history are retained, including unsuccessful episodes.
+
+Local `task_index` values 0–3 and `episode_index` values 0–29 belong to this subset. They must not be joined directly to the independent indices of the training subsets. No prior episode or collection-source identifiers are included.
+
+<a id="metric-coverage"></a>
+
+## Which trajectories evaluate which metrics?
+
+[`annotations/metric_membership.json`](annotations/metric_membership.json) is the **authoritative metric membership**, even when other annotation fields exist on an episode. The same episode can participate in several metrics, so the counts below must not be added to obtain the dataset size.
+
+| Metric / code key | What it evaluates | Eligible trajectories | Episodes / groups | Supporting annotations |
+|---|---|---|---|---|
+| TGA-CT / `tga_easy` | Distinguishing the correct task instruction from other tasks | Desk and flower ED | 4 episodes | `tasks.json`, `task_candidates.json` (14 candidate instructions) |
+| TGA-CF / `tga_hard` | Distinguishing the task goal from counterfactual instructions | Desk and flower ED | 4 episodes | `tasks.json` (4 counterfactual instructions per regular task) |
+| VOC / `voc` | Value ordering along forward task progress | Desk and flower ED | 4 episodes | Subtask intervals and sampling anchors in `episodes.jsonl` |
+| Cycle-VOC / `cycle_voc` | Value ordering along forward and reverse task progress | Desk and flower ED | 4 episodes | Complete trajectories, subtask intervals, and sampling anchors in `episodes.jsonl` |
+| VS / `vs` | Stability of the forward value curve | Desk and flower ED | 4 episodes | The same forward trajectories and sampling basis used by Cycle-VOC |
+| SA / `sa` | Distinguishing terminal task success from failure | All desk and flower episodes | 24 episodes: 16 successful, 8 unsuccessful | `success` in `episodes.jsonl` |
+| SIA / `sia` | Identifying the semantic subtask being performed | Desk and flower ED/MS | 12 episodes | Subtask descriptions in `tasks.json` and local intervals in `episodes.jsonl` |
+| CSVC / `csvc` | Consistency of semantic value increments across successful solutions | Desk and flower ED/MS | 12 episodes / 10 solution groups | `csvc_solutions.json`: groups, semantic/object order, and subtask intervals |
+| TRR / `trr` | Distinguishing effective recovery, ineffective recovery, and error continuation | Desk and flower ER/IR/EC | 12 episodes / 4 three-role groups | `trr_groups.json` and failure/recovery stage intervals in `episodes.jsonl` |
+| FPL / `fpl` | Locating the onset of the target error | Selected desk and flower ER/EC | 8 episodes: 4 ER, 4 EC | `fpl_annotations.json`: target subtask, failure interval, and `failure_onset_frame` |
+| VOC-MEM / `voc_mem` | Value ordering for repeated-action counting using complete history | Drum and button MEM | 6 episodes, all successful | Counting subtask intervals and sampling anchors in `episodes.jsonl` |
+
+All four ED episodes are `0, 1, 12, 14`. Counting episodes are `24–29`. For a per-episode list, inspect [`episode_catalog.parquet`](episode_catalog.parquet), whose `metrics` column is derived from the membership file.
+
+| Trajectory type | Meaning | Count | Applicable metrics in this release |
+|---|---|---:|---|
+| ED | Standard successful demonstration | 4 | TGA-CT, TGA-CF, VOC, Cycle-VOC, VS, SA, SIA, CSVC |
+| MS | Successful alternative solution | 8 | SA, SIA, CSVC |
+| ER | Effective recovery | 4 | SA, TRR, FPL |
+| IR | Ineffective recovery | 4 | SA, TRR |
+| EC | Error continuation | 4 | SA, TRR, FPL |
+| MEM | Complete counting-history episode | 6 | VOC-MEM |
+
+FPL's eight episodes all contain a target error, but the four ER episodes ultimately succeed; they are not eight terminal failures. Each TRR group has one ER, one IR, and one EC member; their roles correspond to `r_plus`, `r_zero`, and `r_minus`. Each regular task has five CSVC solution groups: one standard-repeat group containing two ED instances, and four alternative groups containing one MS instance each. This is **10 groups and 12 instances**, not 12 distinct solutions or exhaustive solution coverage. The six MEM episodes provide successful counting histories; counting failures and counting recovery are not covered.
+
+## Files and annotation semantics
+
+```text
+meta/info.json                         Native format, features, FPS, and path templates
+meta/tasks.parquet                     English instructions indexed by task_index
+meta/episodes/chunk-000/file-000.parquet Native episode lengths and video references
+data/chunk-000/file-*.parquet           Native frame-reference records
+videos/observation.images.top_head/    Head-camera videos
+videos/observation.images.hand_left/   Left-wrist videos
+videos/observation.images.hand_right/  Right-wrist videos
+annotations/                          Separate English evaluation labels
+episode_catalog.parquet               One browsing/index row per episode
+checksums.sha256                       Release-file integrity manifest
+```
+
+Each native frame record contains only `timestamp`, `frame_index`, `episode_index`, `index`, and `task_index`. They locate frames and instructions; they are not model observations. Timestamps are seconds relative to the episode start. This visual export contains no robot state/action vectors, reserved robot interfaces, collection timestamps, collection software versions, or source identities. `codebase_version: v3.0` identifies the native format, and `robot_type` is null. Numeric normalization statistics are omitted because there are no numeric observation features.
+
+Video container/encoder metadata and HEVC supplemental messages were removed by stream copying. Frames were not re-encoded, cropped, resampled, or reordered; decoded pixels and frame timing were preserved. This release copies the final visual-input export unchanged.
+
+| Annotation file | Contents |
+|---|---|
+| [`tasks.json`](annotations/tasks.json) | Four English task instructions, semantic subtask IDs/descriptions, and available counterfactual instructions |
+| [`episodes.jsonl`](annotations/episodes.jsonl) | Thirty success flags, trajectory types, complete per-episode annotation objects, action segments, keyframes, and available flower identities |
+| [`trr_groups.json`](annotations/trr_groups.json) | Four three-role groups and their target subtasks |
+| [`fpl_annotations.json`](annotations/fpl_annotations.json) | Eight selected target errors, target subtasks, and error-onset frames |
+| [`csvc_solutions.json`](annotations/csvc_solutions.json) | Solution groups, semantic/object order, and per-instance subtask intervals |
+| [`metric_membership.json`](annotations/metric_membership.json) | Exact episode/task cohorts for the 11 metric entry points |
+| [`task_candidates.json`](annotations/task_candidates.json) | Fourteen English task candidates for cross-task goal assessment |
+
+All annotation frame indices are local and zero-based. Evaluation intervals use **half-open** `[start_frame, end_frame_exclusive)` ranges. Existing action/keyframe `end_frame_index` endpoint values are retained as supplied; do not silently convert their convention. Semantic subtask IDs describe subtasks, and `group_index`/`solution_index` describe evaluation relationships rather than collection sources.
+
+The per-episode `annotations` object retains available `subtask_segments`, `fpl`, `trr`, `voc`, and `voc_mem` entries. TRR stage keys include `failure_span`, `recovery_span`, `recovery_success_span`, `recovery_failed_span`, and `continue_span`, as applicable. VOC/VOC-MEM entries are **sampling anchors**, not ground-truth value/progress scores or model predictions. Keyframe `label` is an event annotation; `frame_type_name` retains the existing marker category. Flower identities are auxiliary object labels; ordinal flower subtasks keep their existing semantics. Missing detailed physical error labels are not invented.
+
+Annotations are read explicitly and are not automatically added to native LeRobot samples. The presence of a `voc` field on MS/ER/IR/EC does not expand their formal VOC/Cycle-VOC/VS membership; an IR episode is not in the formal FPL cohort.
+
+## Trajectory catalog and HF preview
+
+The repository's `real-world-test-30` configuration exposes a **`test` split with 30 episode rows**, explicitly selecting `episode_catalog.parquet`. Auxiliary JSON reports and navigation documents are not preview datasets. This episode catalog is a browsing index, while LeRobot provides the native per-time-step samples. [`TRAJECTORIES.md`](TRAJECTORIES.md) also lists all 30 episodes, their eligible metrics, and direct links to the three camera videos.
+
+| Field | Type | Meaning |
+|---|---|---|
+| `episode_index`, `task_index` | `int64` | Public indices local to this subset |
+| `task_name`, `instruction` | `string` | English display task name and original English instruction |
+| `trajectory_type` | `string` | ED, MS, ER, IR, EC, or MEM |
+| `success` | `bool` | Terminal task-success annotation |
+| `num_frames` | `int64` | Episode time-step count, counted once across cameras |
+| `duration_s` | `float64` | Episode duration (`num_frames / fps`) |
+| `video_primary`, `video_left_wrist`, `video_right_wrist` | `string` | Video-file paths relative to this subset root |
+| `metrics` | `list[string]` | Eligible metric code keys from `metric_membership.json` |
+
+The catalog's video columns are file references, not decoded model observations. Path references and labels are for navigation/evaluation, not inputs to the value model.
+
+## Download and load
+
+The example is pinned to `578593ef28ee7a2d861ebf38162c928291b0d4a9`, which contains both public test sets, and downloads only this subset. Download and load using the same fixed revision.
+
+```python
+from pathlib import Path
+from huggingface_hub import snapshot_download
+
+repo_id = "wzlllong/value-model-benchmark-training-data"
+prefix = "evaluation/real-world/test-30"
+revision = "578593ef28ee7a2d861ebf38162c928291b0d4a9"  # Fixed data revision containing both test domains.
+snapshot = snapshot_download(
+    repo_id=repo_id,
+    repo_type="dataset",
+    revision=revision,
+    allow_patterns=[f"{prefix}/**"],
+    local_dir="./robovalue-data",
+)
+root = Path(snapshot) / prefix
+```
+
+For an extracted local upload package, use `root = Path("upload/evaluation/real-world/test-30")`. In either case, `root` must contain `meta/`, `data/`, and `videos/`. Install LeRobot with its PyAV video backend, then load the complete native subset:
+
+```python
+from lerobot.datasets.lerobot_dataset import LeRobotDataset
+
+dataset = LeRobotDataset(
+    repo_id="wzlllong/value-model-benchmark-training-data",
+    root=root,
+    video_backend="pyav",
+    tolerance_s=1e-4,
+)
+item = dataset[0]
+camera_keys = [
+    "observation.images.top_head",
+    "observation.images.hand_left",
+    "observation.images.hand_right",
+]
+value_input = {
+    "images": {key: item[key] for key in camera_keys},
+    "instruction": item["task"],
+}
+print(dataset.num_episodes, dataset.num_frames)  # 30, 27498
+```
+
+Loaded camera images use CHW order: head `[3, 400, 640]`, wrists `[3, 528, 640]`. Send **only the selected camera images and instruction** to the value model. Success flags, trajectory types, subtask/solution/group labels, indices, timestamps, and file paths are not model observations. History-based models can request earlier frames through LeRobot's `delta_timestamps`; retain their episode alignment and ordering.
+
+Read labels and metric eligibility separately using the same local episode index:
+
+```python
+import json
+
+episode_index = int(item["episode_index"])
+episodes = {
+    row["episode_index"]: row
+    for row in (
+        json.loads(line)
+        for line in (root / "annotations/episodes.jsonl").read_text().splitlines()
+        if line.strip()
+    )
+}
+membership = json.loads(
+    (root / "annotations/metric_membership.json").read_text()
+)["metrics"]
+labels = episodes[episode_index]
+eligible_metrics = [
+    key for key, cohort in membership.items()
+    if episode_index in cohort["episode_indices"]
+]
+assert int(item["task_index"]) == labels["task_index"]
+```
+
+From the subset directory, verify the release-file hashes with:
+
+```bash
+shasum -a 256 -c checksums.sha256
+```

evaluation/real-world/test-30/README_zh.md

@@ -0,0 +1,183 @@
+# RoboValue 小规模真机公开测试集
+
+[English](README.md) · [仓库概览](../../../README_zh.md) · [查看全部30条轨迹](TRAJECTORIES.md) · [Parquet目录](episode_catalog.parquet)
+
+这是一个**真实世界 In-Domain 诊断测试子集**，包含 **4 个任务、30 条完整轨迹**，涵盖标准示范、多种成功方案、错误、恢复和完整计数历史。数据提供图像、英语任务指令及独立评测标注，支持 **11 个指标入口**，便于检查指标行为和开展小规模评测。它的任务与方案覆盖范围小于完整 RoboValue 排行榜测试集。
+
+本子集位于 `evaluation/real-world/test-30/`，与仓库 `data/` 下的训练示范分区存放，覆盖已有的4个真机任务类别；评测范围以下方各指标的正式轨迹集合为准。为兼容 LeRobot 容器，原生元数据保留 `train: 0:30`；本次发布的用途及 HF 逐轨迹目录划分均为 **test（测试）**。
+
+## 任务与轨迹
+
+| 子集内任务编号 | 任务 | 轨迹数 | 子集内轨迹编号 |
+|---:|---|---:|---|
+| 0 | 整理桌面工作区（Organize Desktop Workstation） | 12 | 0–11 |
+| 1 | 插花与摆放藤条香薰（Arrange Flowers and Reed Diffuser） | 12 | 12–23 |
+| 2 | 敲鼓五次（Strike Drum Five Times） | 3 | 24–26 |
+| 3 | 按按钮五次（Press Button Five Times） | 3 | 27–29 |
+
+数据包含 **27,498 个同步时间步，30 FPS，90 个 HEVC 视频**：每条轨迹各有头部、左手腕、右手腕三个同步视角。时间步按轨迹计数一次，不按机位重复累加。头部视频尺寸为 **640 × 400**，两路腕部视频均为 **640 × 528**。视觉输入源目录约 **382 MB（0.38 GB，十进制）**；本次发布另增加少量说明和目录文件。共 **22 条最终成功、8 条最终失败**；所有轨迹与历史均完整保留，包括失败轨迹。
+
+`task_index` 0–3 和 `episode_index` 0–29 都是本子集内编号，不能直接与训练子集的独立编号关联。文件中不包含原轨迹编号或采集来源标识。
+
+<a id="metric-coverage"></a>
+
+## 哪些轨迹用于哪些指标？
+
+[`annotations/metric_membership.json`](annotations/metric_membership.json) 是**唯一权威的指标归属表**。即使某条轨迹还含有其他标注字段，也应按该文件筛选正式评测集合。同一轨迹可参加多个指标，不能把下表的轨迹数相加作为数据集总量。
+
+| 指标／代码键 | 评测用途 | 适用轨迹 | 轨迹数／组数 | 对应标注 |
+|---|---|---|---|---|
+| TGA-CT／`tga_easy` | 从不同任务的指令中辨别正确任务目标 | 书桌、花艺 ED | 4条 | `tasks.json`、`task_candidates.json`（14个候选指令） |
+| TGA-CF／`tga_hard` | 区分原任务目标与反事实任务指令 | 书桌、花艺 ED | 4条 | `tasks.json`（每个常规任务4条反事实指令） |
+| VOC／`voc` | 检查正向任务进展的价值顺序 | 书桌、花艺 ED | 4条 | `episodes.jsonl` 中的子任务区间和采样锚点 |
+| Cycle-VOC／`cycle_voc` | 检查正向及反向播放下的进展价值顺序 | 书桌、花艺 ED | 4条 | 完整轨迹、子任务区间、采样锚点 |
+| VS／`vs` | 检查正向价值曲线的稳定性 | 书桌、花艺 ED | 4条 | 与 Cycle-VOC 共享的正向轨迹和采样依据 |
+| SA／`sa` | 区分最终任务成功与失败 | 书桌、花艺全部轨迹 | 24条：16成功、8失败 | `episodes.jsonl` 中的 `success` |
+| SIA／`sia` | 识别当前执行的语义子任务 | 书桌、花艺 ED/MS | 12条 | `tasks.json` 的子任务描述及 `episodes.jsonl` 的局部区间 |
+| CSVC／`csvc` | 检查不同成功方案中的语义价值增量一致性 | 书桌、花艺 ED/MS | 12条／10个方案组 | `csvc_solutions.json` 中的分组、语义／物体顺序及子任务区间 |
+| TRR／`trr` | 区分有效恢复、无效恢复和错误延续 | 书桌、花艺 ER/IR/EC | 12条／4个三角色组 | `trr_groups.json` 及 `episodes.jsonl` 的失败／恢复阶段区间 |
+| FPL／`fpl` | 定位目标错误的发生起点 | 选定的书桌、花艺 ER/EC | 8条：4 ER、4 EC | `fpl_annotations.json` 中的目标子任务、错误区间、`failure_onset_frame` |
+| VOC-MEM／`voc_mem` | 利用完整历史判断重复动作的计数进展顺序 | 敲鼓、按钮 MEM | 6条，全部最终成功 | `episodes.jsonl` 中的计数子任务区间和采样锚点 |
+
+4条 ED 的编号为 `0、1、12、14`；计数轨迹的编号为 `24–29`。每条轨迹的适用指标也可在 [`episode_catalog.parquet`](episode_catalog.parquet) 的 `metrics` 列查看，该列由权威归属表反向生成。
+
+| 轨迹类型 | 含义 | 数量 | 本次发布适用指标 |
+|---|---|---:|---|
+| ED | 标准成功示范 | 4 | TGA-CT、TGA-CF、VOC、Cycle-VOC、VS、SA、SIA、CSVC |
+| MS | 成功的替代方案 | 8 | SA、SIA、CSVC |
+| ER | 有效恢复 | 4 | SA、TRR、FPL |
+| IR | 无效恢复 | 4 | SA、TRR |
+| EC | 错误延续 | 4 | SA、TRR、FPL |
+| MEM | 保留完整历史的计数轨迹 | 6 | VOC-MEM |
+
+FPL 的8条均含目标错误，但4条 ER 最终恢复成功，不能把它们称为8条最终失败轨迹。每个 TRR 组包含 ER、IR、EC 各一条，对应 `r_plus`、`r_zero`、`r_minus`。CSVC 每个常规任务有5个方案组：1个标准重复组含2条 ED，另4个替代方案组各含1条 MS；合计 **10个方案组、12个实例**，不是12种不同解法，也不是所有可能方案。6条 MEM 当前全部成功，覆盖成功计数历史，尚未覆盖计数失败或计数恢复。
+
+## 文件与标注语义
+
+```text
+meta/info.json                         原生格式、字段、FPS与路径模板
+meta/tasks.parquet                     按task_index索引的英语指令
+meta/episodes/chunk-000/file-000.parquet 原生轨迹长度与视频引用
+data/chunk-000/file-*.parquet           原生逐帧定位记录
+videos/observation.images.top_head/    头部视频
+videos/observation.images.hand_left/   左手腕视频
+videos/observation.images.hand_right/  右手腕视频
+annotations/                          独立英语评测标注
+episode_catalog.parquet               每条轨迹一行的浏览索引
+checksums.sha256                       发布文件完整性清单
+```
+
+原生逐帧记录仅含 `timestamp`、`frame_index`、`episode_index`、`index` 和 `task_index`，用于定位帧和指令，不是模型观测。`timestamp` 是从轨迹开始计算的秒数。视觉导出不含机器人状态／动作向量、预留机器人接口、采集时间、采集软件版本或来源标识。`codebase_version: v3.0` 是原生格式标识，`robot_type` 为 null；因不存在数值观测特征，未生成数值归一化统计。
+
+视频容器／编码器元数据和 HEVC 补充消息已通过流复制移除。图像未经重新编码、裁剪、重采样或重排，解码像素和帧时间轴均保留。本次整理直接复制最终视觉输入导出，数据内容不变。
+
+| 标注文件 | 内容 |
+|---|---|
+| [`tasks.json`](annotations/tasks.json) | 4个英语任务指令、语义子任务编号／描述及已有反事实指令 |
+| [`episodes.jsonl`](annotations/episodes.jsonl) | 30条成功标记、轨迹类型、完整逐轨迹标注、动作片段、关键帧及已有花朵身份标注 |
+| [`trr_groups.json`](annotations/trr_groups.json) | 4个三角色分组及目标子任务 |
+| [`fpl_annotations.json`](annotations/fpl_annotations.json) | 8个选定目标错误、目标子任务及错误起点 |
+| [`csvc_solutions.json`](annotations/csvc_solutions.json) | 方案分组、语义／物体顺序及各实例的子任务区间 |
+| [`metric_membership.json`](annotations/metric_membership.json) | 11个指标入口的精确任务／轨迹集合 |
+| [`task_candidates.json`](annotations/task_candidates.json) | 14个用于跨任务目标辨别的英语候选任务指令 |
+
+所有标注帧编号均为本轨迹内从0开始的局部编号。评测区间采用**左闭右开** `[start_frame, end_frame_exclusive)`；已有动作／关键帧 `end_frame_index` 的端点值按原标注保留，不应自行转换约定。语义子任务编号用于描述子任务；`group_index` 和 `solution_index` 用于表达评测关系，不是采集来源标识。
+
+逐轨迹 `annotations` 对象保留已有 `subtask_segments`、`fpl`、`trr`、`voc` 和 `voc_mem` 字段。TRR 阶段区间视轨迹包含 `failure_span`、`recovery_span`、`recovery_success_span`、`recovery_failed_span`、`continue_span`。VOC／VOC-MEM 条目是**采样锚点**，不是逐帧真实价值／进度分数或模型预测。关键帧 `label` 为事件标注，`frame_type_name` 保留已有标记类别。花朵身份为辅助物体标注，按第几朵定义的子任务语义保持不变；缺失的详细物理错误标签不会被补造。
+
+标注需显式读取，LeRobot 不会自动把它们加入 sample。MS/ER/IR/EC 即使含有 `voc` 字段，也不属于本次正式 VOC／Cycle-VOC／VS 集合；IR 不属于正式 FPL 集合。
+
+## 逐轨迹目录与 HF 预览
+
+仓库的 `real-world-test-30` 配置显式指向 `episode_catalog.parquet`，提供 **30行、划分为 `test` 的轨迹预览**。辅助验证 JSON 和导航说明不作为预览数据。目录每行代表一条轨迹；实际逐时间步样本由 LeRobot 读取。[`TRAJECTORIES.md`](TRAJECTORIES.md) 另提供30条轨迹的逐条表格、适用指标及三个机位的视频链接。
+
+| 字段 | 类型 | 含义 |
+|---|---|---|
+| `episode_index`、`task_index` | `int64` | 本子集内的公开编号 |
+| `task_name`、`instruction` | `string` | 英语展示任务名与原英语任务指令 |
+| `trajectory_type` | `string` | ED、MS、ER、IR、EC或MEM |
+| `success` | `bool` | 最终任务成功标记 |
+| `num_frames` | `int64` | 轨迹时间步数，三个机位只计一次 |
+| `duration_s` | `float64` | 轨迹时长（`num_frames / fps`） |
+| `video_primary`、`video_left_wrist`、`video_right_wrist` | `string` | 相对于本子集根目录的视频文件路径 |
+| `metrics` | `list[string]` | 由 `metric_membership.json` 提供的指标代码键 |
+
+目��里的视频列是文件引用，不是已解码的模型观测。文件路径及标签用于浏览和评测，不作为价值模型输入。
+
+## 单独下载与加载
+
+以下示例固定到包含两个公开测试集的版本 `578593ef28ee7a2d861ebf38162c928291b0d4a9`，仅下载本子集。下载和加载使用同一提交。
+
+```python
+from pathlib import Path
+from huggingface_hub import snapshot_download
+
+repo_id = "wzlllong/value-model-benchmark-training-data"
+prefix = "evaluation/real-world/test-30"
+revision = "578593ef28ee7a2d861ebf38162c928291b0d4a9"  # 包含两域测试集的固定发布版本。
+snapshot = snapshot_download(
+    repo_id=repo_id,
+    repo_type="dataset",
+    revision=revision,
+    allow_patterns=[f"{prefix}/**"],
+    local_dir="./robovalue-data",
+)
+root = Path(snapshot) / prefix
+```
+
+若使用解压后的本地上传包，设置 `root = Path("upload/evaluation/real-world/test-30")`。`root` 应直接包含 `meta/`、`data/` 和 `videos/`。安装带 PyAV 视频后端的 LeRobot 后，读取完整原生数据集：
+
+```python
+from lerobot.datasets.lerobot_dataset import LeRobotDataset
+
+dataset = LeRobotDataset(
+    repo_id="wzlllong/value-model-benchmark-training-data",
+    root=root,
+    video_backend="pyav",
+    tolerance_s=1e-4,
+)
+item = dataset[0]
+camera_keys = [
+    "observation.images.top_head",
+    "observation.images.hand_left",
+    "observation.images.hand_right",
+]
+value_input = {
+    "images": {key: item[key] for key in camera_keys},
+    "instruction": item["task"],
+}
+print(dataset.num_episodes, dataset.num_frames)  # 30, 27498
+```
+
+加载后的图像使用 CHW 顺序：头部 `[3, 400, 640]`，腕部 `[3, 528, 640]`。价值模型仅接收**所选机位图像和指令**。成功标记、轨迹类型、子任务／方案／分组标签、编号、时间戳及文件路径均不是模型观测。需要历史的模型可使用 LeRobot `delta_timestamps` 读取更早帧，并保持轨迹内的时间顺序和对齐。
+
+通过同一局部轨迹编号，单独读取标签和指标归属：
+
+```python
+import json
+
+episode_index = int(item["episode_index"])
+episodes = {
+    row["episode_index"]: row
+    for row in (
+        json.loads(line)
+        for line in (root / "annotations/episodes.jsonl").read_text().splitlines()
+        if line.strip()
+    )
+}
+membership = json.loads(
+    (root / "annotations/metric_membership.json").read_text()
+)["metrics"]
+labels = episodes[episode_index]
+eligible_metrics = [
+    key for key, cohort in membership.items()
+    if episode_index in cohort["episode_indices"]
+]
+assert int(item["task_index"]) == labels["task_index"]
+```
+
+在子集目录中运行以下命令检查发布文件哈希：
+
+```bash
+shasum -a 256 -c checksums.sha256
+```

evaluation/real-world/test-30/TRAJECTORIES.md

@@ -0,0 +1,52 @@
+# Real-World Test-30: Trajectory Index / 真机30条轨迹索引
+
+[English documentation](README.md) · [中文说明](README_zh.md)
+
+One row represents one complete trajectory. Three video links refer to its synchronized camera views.
+每行是一条完整轨迹，三个视频链接是同一轨迹的同步机位。指标列表仅由正式 `annotations/metric_membership.json` 生成。
+
+Episode/task indices are local to this subset; they are independent of the training data and the other test domain.
+轨迹和任务编号仅在本子集中有效，与训练集及另一测试领域的编号独立。
+
+| Task index | Task / 任务 | Episodes / 轨迹数 |
+| ---: | --- | ---: |
+| 0 | Organize Desktop Workstation / 整理书桌 | 12 |
+| 1 | Arrange Flowers and Reed Diffuser / 花艺与香薰 | 12 |
+| 2 | Strike Drum Five Times / 敲鼓五次 | 3 |
+| 3 | Press Button Five Times / 按按钮五次 | 3 |
+
+| Episode | Task | Type | Final success / 最终成功 | Frames | Duration (s) | Head / 头部 | Left wrist / 左腕 | Right wrist / 右腕 | Metrics / 指标 |
+| ---: | ---: | --- | --- | ---: | ---: | --- | --- | --- | --- |
+| 0 | 0 | ED | True / 是 | 1029 | 34.300 | [Video](videos/observation.images.top_head/chunk-000/file-000.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-000.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-000.mp4) | TGA-CT, TGA-CF, VOC, Cycle-VOC, VS, SA, SIA, CSVC |
+| 1 | 0 | ED | True / 是 | 1045 | 34.833 | [Video](videos/observation.images.top_head/chunk-000/file-001.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-001.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-001.mp4) | TGA-CT, TGA-CF, VOC, Cycle-VOC, VS, SA, SIA, CSVC |
+| 2 | 0 | MS | True / 是 | 998 | 33.267 | [Video](videos/observation.images.top_head/chunk-000/file-002.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-002.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-002.mp4) | SA, SIA, CSVC |
+| 3 | 0 | MS | True / 是 | 892 | 29.733 | [Video](videos/observation.images.top_head/chunk-000/file-003.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-003.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-003.mp4) | SA, SIA, CSVC |
+| 4 | 0 | MS | True / 是 | 818 | 27.267 | [Video](videos/observation.images.top_head/chunk-000/file-004.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-004.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-004.mp4) | SA, SIA, CSVC |
+| 5 | 0 | MS | True / 是 | 982 | 32.733 | [Video](videos/observation.images.top_head/chunk-000/file-005.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-005.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-005.mp4) | SA, SIA, CSVC |
+| 6 | 0 | ER | True / 是 | 1146 | 38.200 | [Video](videos/observation.images.top_head/chunk-000/file-006.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-006.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-006.mp4) | SA, TRR, FPL |
+| 7 | 0 | IR | False / 否 | 1161 | 38.700 | [Video](videos/observation.images.top_head/chunk-000/file-007.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-007.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-007.mp4) | SA, TRR |
+| 8 | 0 | EC | False / 否 | 906 | 30.200 | [Video](videos/observation.images.top_head/chunk-000/file-008.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-008.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-008.mp4) | SA, TRR, FPL |
+| 9 | 0 | EC | False / 否 | 817 | 27.233 | [Video](videos/observation.images.top_head/chunk-000/file-009.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-009.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-009.mp4) | SA, TRR, FPL |
+| 10 | 0 | IR | False / 否 | 937 | 31.233 | [Video](videos/observation.images.top_head/chunk-000/file-010.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-010.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-010.mp4) | SA, TRR |
+| 11 | 0 | ER | True / 是 | 922 | 30.733 | [Video](videos/observation.images.top_head/chunk-000/file-011.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-011.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-011.mp4) | SA, TRR, FPL |
+| 12 | 1 | ED | True / 是 | 934 | 31.133 | [Video](videos/observation.images.top_head/chunk-000/file-012.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-012.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-012.mp4) | TGA-CT, TGA-CF, VOC, Cycle-VOC, VS, SA, SIA, CSVC |
+| 13 | 1 | EC | False / 否 | 905 | 30.167 | [Video](videos/observation.images.top_head/chunk-000/file-013.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-013.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-013.mp4) | SA, TRR, FPL |
+| 14 | 1 | ED | True / 是 | 939 | 31.300 | [Video](videos/observation.images.top_head/chunk-000/file-014.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-014.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-014.mp4) | TGA-CT, TGA-CF, VOC, Cycle-VOC, VS, SA, SIA, CSVC |
+| 15 | 1 | MS | True / 是 | 1048 | 34.933 | [Video](videos/observation.images.top_head/chunk-000/file-015.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-015.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-015.mp4) | SA, SIA, CSVC |
+| 16 | 1 | IR | False / 否 | 1081 | 36.033 | [Video](videos/observation.images.top_head/chunk-000/file-016.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-016.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-016.mp4) | SA, TRR |
+| 17 | 1 | EC | False / 否 | 856 | 28.533 | [Video](videos/observation.images.top_head/chunk-000/file-017.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-017.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-017.mp4) | SA, TRR, FPL |
+| 18 | 1 | MS | True / 是 | 1032 | 34.400 | [Video](videos/observation.images.top_head/chunk-000/file-018.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-018.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-018.mp4) | SA, SIA, CSVC |
+| 19 | 1 | MS | True / 是 | 936 | 31.200 | [Video](videos/observation.images.top_head/chunk-000/file-019.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-019.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-019.mp4) | SA, SIA, CSVC |
+| 20 | 1 | ER | True / 是 | 1265 | 42.167 | [Video](videos/observation.images.top_head/chunk-000/file-020.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-020.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-020.mp4) | SA, TRR, FPL |
+| 21 | 1 | IR | False / 否 | 1130 | 37.667 | [Video](videos/observation.images.top_head/chunk-000/file-021.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-021.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-021.mp4) | SA, TRR |
+| 22 | 1 | MS | True / 是 | 1002 | 33.400 | [Video](videos/observation.images.top_head/chunk-000/file-022.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-022.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-022.mp4) | SA, SIA, CSVC |
+| 23 | 1 | ER | True / 是 | 1291 | 43.033 | [Video](videos/observation.images.top_head/chunk-000/file-023.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-023.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-023.mp4) | SA, TRR, FPL |
+| 24 | 2 | MEM | True / 是 | 575 | 19.167 | [Video](videos/observation.images.top_head/chunk-000/file-024.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-024.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-024.mp4) | VOC-MEM |
+| 25 | 2 | MEM | True / 是 | 560 | 18.667 | [Video](videos/observation.images.top_head/chunk-000/file-025.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-025.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-025.mp4) | VOC-MEM |
+| 26 | 2 | MEM | True / 是 | 590 | 19.667 | [Video](videos/observation.images.top_head/chunk-000/file-026.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-026.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-026.mp4) | VOC-MEM |
+| 27 | 3 | MEM | True / 是 | 572 | 19.067 | [Video](videos/observation.images.top_head/chunk-000/file-027.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-027.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-027.mp4) | VOC-MEM |
+| 28 | 3 | MEM | True / 是 | 587 | 19.567 | [Video](videos/observation.images.top_head/chunk-000/file-028.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-028.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-028.mp4) | VOC-MEM |
+| 29 | 3 | MEM | True / 是 | 542 | 18.067 | [Video](videos/observation.images.top_head/chunk-000/file-029.mp4) | [Video](videos/observation.images.hand_left/chunk-000/file-029.mp4) | [Video](videos/observation.images.hand_right/chunk-000/file-029.mp4) | VOC-MEM |
+
+Duration is frame count divided by FPS, counted once per episode. Files and evaluation labels are for navigation and scoring; model observations are images and the instruction.
+时长为帧数除以帧率，每条轨迹只计一次。文件路径和评测标签用于浏览与评分；模型观测为图像和指令。

evaluation/real-world/test-30/annotations/csvc_solutions.json

@@ -0,0 +1,678 @@
+{
+  "tasks": [
+    {
+      "task_index": 0,
+      "solutions": [
+        {
+          "solution_index": 0,
+          "episode_indices": [
+            0,
+            1
+          ],
+          "semantic_order": [
+            "001",
+            "002",
+            "003",
+            "004",
+            "005"
+          ],
+          "object_order": [
+            "pen",
+            "eraser",
+            "drawer",
+            "lamp_switch",
+            "lamp_plug"
+          ],
+          "solution_kind": "standard_repeat_group",
+          "instances": [
+            {
+              "episode_index": 0,
+              "subtask_segments": [
+                {
+                  "id": "001",
+                  "start_frame": 19,
+                  "end_frame_exclusive": 199,
+                  "object_id": "pen"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 199,
+                  "end_frame_exclusive": 329,
+                  "object_id": "eraser"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 329,
+                  "end_frame_exclusive": 493,
+                  "object_id": "drawer"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 493,
+                  "end_frame_exclusive": 633,
+                  "object_id": "lamp_switch"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 633,
+                  "end_frame_exclusive": 979,
+                  "object_id": "lamp_plug"
+                }
+              ],
+              "execution_hand": "right"
+            },
+            {
+              "episode_index": 1,
+              "subtask_segments": [
+                {
+                  "id": "001",
+                  "start_frame": 24,
+                  "end_frame_exclusive": 219,
+                  "object_id": "pen"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 219,
+                  "end_frame_exclusive": 363,
+                  "object_id": "eraser"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 363,
+                  "end_frame_exclusive": 489,
+                  "object_id": "drawer"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 489,
+                  "end_frame_exclusive": 638,
+                  "object_id": "lamp_switch"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 638,
+                  "end_frame_exclusive": 971,
+                  "object_id": "lamp_plug"
+                }
+              ],
+              "execution_hand": "right"
+            }
+          ]
+        },
+        {
+          "solution_index": 1,
+          "episode_indices": [
+            2
+          ],
+          "semantic_order": [
+            "001",
+            "004",
+            "005",
+            "002",
+            "003"
+          ],
+          "object_order": [
+            "pen",
+            "lamp_switch",
+            "lamp_plug",
+            "eraser",
+            "drawer"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 2,
+              "subtask_segments": [
+                {
+                  "id": "001",
+                  "start_frame": 20,
+                  "end_frame_exclusive": 196,
+                  "object_id": "pen"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 196,
+                  "end_frame_exclusive": 354,
+                  "object_id": "lamp_switch"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 354,
+                  "end_frame_exclusive": 668,
+                  "object_id": "lamp_plug"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 668,
+                  "end_frame_exclusive": 780,
+                  "object_id": "eraser"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 780,
+                  "end_frame_exclusive": 972,
+                  "object_id": "drawer"
+                }
+              ],
+              "execution_hand": "right"
+            }
+          ]
+        },
+        {
+          "solution_index": 2,
+          "episode_indices": [
+            4
+          ],
+          "semantic_order": [
+            "004",
+            "002",
+            "003",
+            "001",
+            "005"
+          ],
+          "object_order": [
+            "lamp_switch",
+            "eraser",
+            "drawer",
+            "pen",
+            "lamp_plug"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 4,
+              "subtask_segments": [
+                {
+                  "id": "004",
+                  "start_frame": 16,
+                  "end_frame_exclusive": 139,
+                  "object_id": "lamp_switch"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 139,
+                  "end_frame_exclusive": 228,
+                  "object_id": "eraser"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 228,
+                  "end_frame_exclusive": 333,
+                  "object_id": "drawer"
+                },
+                {
+                  "id": "001",
+                  "start_frame": 333,
+                  "end_frame_exclusive": 523,
+                  "object_id": "pen"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 523,
+                  "end_frame_exclusive": 800,
+                  "object_id": "lamp_plug"
+                }
+              ],
+              "execution_hand": "right"
+            }
+          ]
+        },
+        {
+          "solution_index": 3,
+          "episode_indices": [
+            3
+          ],
+          "semantic_order": [
+            "002",
+            "003",
+            "004",
+            "001",
+            "005"
+          ],
+          "object_order": [
+            "eraser",
+            "drawer",
+            "lamp_switch",
+            "pen",
+            "lamp_plug"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 3,
+              "subtask_segments": [
+                {
+                  "id": "002",
+                  "start_frame": 17,
+                  "end_frame_exclusive": 135,
+                  "object_id": "eraser"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 135,
+                  "end_frame_exclusive": 267,
+                  "object_id": "drawer"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 267,
+                  "end_frame_exclusive": 368,
+                  "object_id": "lamp_switch"
+                },
+                {
+                  "id": "001",
+                  "start_frame": 368,
+                  "end_frame_exclusive": 546,
+                  "object_id": "pen"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 546,
+                  "end_frame_exclusive": 842,
+                  "object_id": "lamp_plug"
+                }
+              ],
+              "execution_hand": "left"
+            }
+          ]
+        },
+        {
+          "solution_index": 4,
+          "episode_indices": [
+            5
+          ],
+          "semantic_order": [
+            "004",
+            "001",
+            "005",
+            "002",
+            "003"
+          ],
+          "object_order": [
+            "lamp_switch",
+            "pen",
+            "lamp_plug",
+            "eraser",
+            "drawer"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 5,
+              "subtask_segments": [
+                {
+                  "id": "004",
+                  "start_frame": 35,
+                  "end_frame_exclusive": 188,
+                  "object_id": "lamp_switch"
+                },
+                {
+                  "id": "001",
+                  "start_frame": 188,
+                  "end_frame_exclusive": 349,
+                  "object_id": "pen"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 349,
+                  "end_frame_exclusive": 647,
+                  "object_id": "lamp_plug"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 647,
+                  "end_frame_exclusive": 749,
+                  "object_id": "eraser"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 749,
+                  "end_frame_exclusive": 928,
+                  "object_id": "drawer"
+                }
+              ],
+              "execution_hand": "left"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "task_index": 1,
+      "solutions": [
+        {
+          "solution_index": 0,
+          "episode_indices": [
+            12,
+            14
+          ],
+          "semantic_order": [
+            "001",
+            "002",
+            "003",
+            "004",
+            "005"
+          ],
+          "object_order": [
+            "yellow_sunflower",
+            "pink_rose",
+            "white_lily",
+            "diffuser_bottle",
+            "diffuser_reed"
+          ],
+          "solution_kind": "standard_repeat_group",
+          "instances": [
+            {
+              "episode_index": 12,
+              "subtask_segments": [
+                {
+                  "id": "001",
+                  "start_frame": 24,
+                  "end_frame_exclusive": 170,
+                  "object_id": "yellow_sunflower"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 170,
+                  "end_frame_exclusive": 343,
+                  "object_id": "pink_rose"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 343,
+                  "end_frame_exclusive": 535,
+                  "object_id": "white_lily"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 535,
+                  "end_frame_exclusive": 704,
+                  "object_id": "diffuser_bottle"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 704,
+                  "end_frame_exclusive": 928,
+                  "object_id": "diffuser_reed"
+                }
+              ],
+              "execution_hand": "dual_arm_subtask_division"
+            },
+            {
+              "episode_index": 14,
+              "subtask_segments": [
+                {
+                  "id": "001",
+                  "start_frame": 28,
+                  "end_frame_exclusive": 168,
+                  "object_id": "yellow_sunflower"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 168,
+                  "end_frame_exclusive": 283,
+                  "object_id": "pink_rose"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 283,
+                  "end_frame_exclusive": 447,
+                  "object_id": "white_lily"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 447,
+                  "end_frame_exclusive": 592,
+                  "object_id": "diffuser_bottle"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 592,
+                  "end_frame_exclusive": 841,
+                  "object_id": "diffuser_reed"
+                }
+              ],
+              "execution_hand": "dual_arm_subtask_division"
+            }
+          ]
+        },
+        {
+          "solution_index": 1,
+          "episode_indices": [
+            19
+          ],
+          "semantic_order": [
+            "005",
+            "004",
+            "001",
+            "002",
+            "003"
+          ],
+          "object_order": [
+            "diffuser_reed",
+            "diffuser_bottle",
+            "white_lily",
+            "pink_rose",
+            "yellow_sunflower"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 19,
+              "subtask_segments": [
+                {
+                  "id": "005",
+                  "start_frame": 37,
+                  "end_frame_exclusive": 245,
+                  "object_id": "diffuser_reed"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 245,
+                  "end_frame_exclusive": 393,
+                  "object_id": "diffuser_bottle"
+                },
+                {
+                  "id": "001",
+                  "start_frame": 393,
+                  "end_frame_exclusive": 557,
+                  "object_id": "white_lily"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 557,
+                  "end_frame_exclusive": 734,
+                  "object_id": "pink_rose"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 734,
+                  "end_frame_exclusive": 908,
+                  "object_id": "yellow_sunflower"
+                }
+              ],
+              "execution_hand": "dual_arm_subtask_division"
+            }
+          ]
+        },
+        {
+          "solution_index": 2,
+          "episode_indices": [
+            22
+          ],
+          "semantic_order": [
+            "001",
+            "002",
+            "003",
+            "005",
+            "004"
+          ],
+          "object_order": [
+            "pink_rose",
+            "white_lily",
+            "yellow_sunflower",
+            "diffuser_reed",
+            "diffuser_bottle"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 22,
+              "subtask_segments": [
+                {
+                  "id": "001",
+                  "start_frame": 12,
+                  "end_frame_exclusive": 167,
+                  "object_id": "pink_rose"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 167,
+                  "end_frame_exclusive": 335,
+                  "object_id": "white_lily"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 335,
+                  "end_frame_exclusive": 563,
+                  "object_id": "yellow_sunflower"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 563,
+                  "end_frame_exclusive": 828,
+                  "object_id": "diffuser_reed"
+                },
+                {
+                  "id": "004",
+                  "start_frame": 828,
+                  "end_frame_exclusive": 984,
+                  "object_id": "diffuser_bottle"
+                }
+              ],
+              "execution_hand": "dual_arm_subtask_division"
+            }
+          ]
+        },
+        {
+          "solution_index": 3,
+          "episode_indices": [
+            18
+          ],
+          "semantic_order": [
+            "004",
+            "005",
+            "001",
+            "002",
+            "003"
+          ],
+          "object_order": [
+            "diffuser_bottle",
+            "diffuser_reed",
+            "pink_rose",
+            "yellow_sunflower",
+            "white_lily"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 18,
+              "subtask_segments": [
+                {
+                  "id": "004",
+                  "start_frame": 27,
+                  "end_frame_exclusive": 167,
+                  "object_id": "diffuser_bottle"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 167,
+                  "end_frame_exclusive": 466,
+                  "object_id": "diffuser_reed"
+                },
+                {
+                  "id": "001",
+                  "start_frame": 466,
+                  "end_frame_exclusive": 625,
+                  "object_id": "pink_rose"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 625,
+                  "end_frame_exclusive": 776,
+                  "object_id": "yellow_sunflower"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 776,
+                  "end_frame_exclusive": 987,
+                  "object_id": "white_lily"
+                }
+              ],
+              "execution_hand": "dual_arm_subtask_division"
+            }
+          ]
+        },
+        {
+          "solution_index": 4,
+          "episode_indices": [
+            15
+          ],
+          "semantic_order": [
+            "004",
+            "001",
+            "002",
+            "003",
+            "005"
+          ],
+          "object_order": [
+            "diffuser_bottle",
+            "yellow_sunflower",
+            "pink_rose",
+            "white_lily",
+            "diffuser_reed"
+          ],
+          "solution_kind": "diverse_solution_group",
+          "instances": [
+            {
+              "episode_index": 15,
+              "subtask_segments": [
+                {
+                  "id": "004",
+                  "start_frame": 10,
+                  "end_frame_exclusive": 183,
+                  "object_id": "diffuser_bottle"
+                },
+                {
+                  "id": "001",
+                  "start_frame": 183,
+                  "end_frame_exclusive": 347,
+                  "object_id": "yellow_sunflower"
+                },
+                {
+                  "id": "002",
+                  "start_frame": 347,
+                  "end_frame_exclusive": 476,
+                  "object_id": "pink_rose"
+                },
+                {
+                  "id": "003",
+                  "start_frame": 476,
+                  "end_frame_exclusive": 677,
+                  "object_id": "white_lily"
+                },
+                {
+                  "id": "005",
+                  "start_frame": 677,
+                  "end_frame_exclusive": 1007,
+                  "object_id": "diffuser_reed"
+                }
+              ],
+              "execution_hand": "dual_arm_subtask_division"
+            }
+          ]
+        }
+      ]
+    }
+  ]
+}

evaluation/real-world/test-30/annotations/fpl_annotations.json

@@ -0,0 +1,108 @@
+{
+  "records": [
+    {
+      "episode_index": 8,
+      "task_index": 0,
+      "trr_group_index": 0,
+      "trajectory_type": "EC",
+      "target_subtask_id": "001",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 194,
+        "end_frame_exclusive": 228
+      },
+      "failure_onset_frame": 194
+    },
+    {
+      "episode_index": 6,
+      "task_index": 0,
+      "trr_group_index": 0,
+      "trajectory_type": "ER",
+      "target_subtask_id": "001",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 196,
+        "end_frame_exclusive": 237
+      },
+      "failure_onset_frame": 196
+    },
+    {
+      "episode_index": 9,
+      "task_index": 0,
+      "trr_group_index": 1,
+      "trajectory_type": "EC",
+      "target_subtask_id": "002",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 199,
+        "end_frame_exclusive": 252
+      },
+      "failure_onset_frame": 199
+    },
+    {
+      "episode_index": 11,
+      "task_index": 0,
+      "trr_group_index": 1,
+      "trajectory_type": "ER",
+      "target_subtask_id": "002",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 199,
+        "end_frame_exclusive": 246
+      },
+      "failure_onset_frame": 199
+    },
+    {
+      "episode_index": 13,
+      "task_index": 1,
+      "trr_group_index": 2,
+      "trajectory_type": "EC",
+      "target_subtask_id": "001",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 78,
+        "end_frame_exclusive": 117
+      },
+      "failure_onset_frame": 78
+    },
+    {
+      "episode_index": 20,
+      "task_index": 1,
+      "trr_group_index": 2,
+      "trajectory_type": "ER",
+      "target_subtask_id": "001",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 76,
+        "end_frame_exclusive": 125
+      },
+      "failure_onset_frame": 76
+    },
+    {
+      "episode_index": 17,
+      "task_index": 1,
+      "trr_group_index": 3,
+      "trajectory_type": "EC",
+      "target_subtask_id": "005",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 750,
+        "end_frame_exclusive": 766
+      },
+      "failure_onset_frame": 750
+    },
+    {
+      "episode_index": 23,
+      "task_index": 1,
+      "trr_group_index": 3,
+      "trajectory_type": "ER",
+      "target_subtask_id": "005",
+      "unique_target_error_count": 1,
+      "failure_span": {
+        "start_frame": 928,
+        "end_frame_exclusive": 964
+      },
+      "failure_onset_frame": 928
+    }
+  ]
+}

evaluation/real-world/test-30/annotations/metric_membership.json

@@ -0,0 +1,186 @@
+{
+  "metrics": {
+    "tga_easy": {
+      "episode_indices": [
+        0,
+        1,
+        12,
+        14
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "tga_hard": {
+      "episode_indices": [
+        0,
+        1,
+        12,
+        14
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "voc": {
+      "episode_indices": [
+        0,
+        1,
+        12,
+        14
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "cycle_voc": {
+      "episode_indices": [
+        0,
+        1,
+        12,
+        14
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "vs": {
+      "episode_indices": [
+        0,
+        1,
+        12,
+        14
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "sa": {
+      "episode_indices": [
+        0,
+        1,
+        2,
+        3,
+        4,
+        5,
+        6,
+        7,
+        8,
+        9,
+        10,
+        11,
+        12,
+        13,
+        14,
+        15,
+        16,
+        17,
+        18,
+        19,
+        20,
+        21,
+        22,
+        23
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "sia": {
+      "episode_indices": [
+        0,
+        1,
+        2,
+        3,
+        4,
+        5,
+        12,
+        14,
+        15,
+        18,
+        19,
+        22
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "trr": {
+      "episode_indices": [
+        6,
+        7,
+        8,
+        11,
+        10,
+        9,
+        13,
+        20,
+        21,
+        16,
+        17,
+        23
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "csvc": {
+      "episode_indices": [
+        0,
+        1,
+        2,
+        4,
+        3,
+        5,
+        12,
+        14,
+        19,
+        22,
+        18,
+        15
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    },
+    "voc_mem": {
+      "episode_indices": [
+        24,
+        25,
+        26,
+        27,
+        28,
+        29
+      ],
+      "task_indices": [
+        2,
+        3
+      ]
+    },
+    "fpl": {
+      "episode_indices": [
+        8,
+        6,
+        9,
+        11,
+        13,
+        20,
+        17,
+        23
+      ],
+      "task_indices": [
+        0,
+        1
+      ]
+    }
+  }
+}

evaluation/real-world/test-30/annotations/task_candidates.json

@@ -0,0 +1,62 @@
+{
+  "candidates": [
+    {
+      "candidate_index": 0,
+      "instruction": "Place the teacup on the cat-head mat, the teapot on the yellow mat, the lid on the teapot, the tongs in their holder, and the canister on the round mat. Complete these placements in any order, provided the teapot is placed before its lid."
+    },
+    {
+      "candidate_index": 1,
+      "instruction": "Place the pink, white, and yellow flowers consecutively in the vase, in any order. Place the diffuser bottle on the coaster and insert the reed stick into it, in either order. Either diffuser action may occur before or after the flower sequence, but neither may occur between flower placements.",
+      "task_index": 1
+    },
+    {
+      "candidate_index": 2,
+      "instruction": "Place the three seasoning jars on the two-tier rack in any order, on either tier and in any positions."
+    },
+    {
+      "candidate_index": 3,
+      "instruction": "Hold the pen holder with its opening upward, insert all four pens one by one in any order, then place the filled holder upright."
+    },
+    {
+      "candidate_index": 4,
+      "instruction": "Insert the three plastic plates into the rack in any order and in any slots."
+    },
+    {
+      "candidate_index": 5,
+      "instruction": "Center and align the closed book in the target rectangle. Place the pen along the spine of the open book, close the book, and stack it on the first book."
+    },
+    {
+      "candidate_index": 6,
+      "instruction": "Open the upper drawer, cap the lipstick, and place the lipstick and eyebrow pencil inside. Close the drawer once both are inside. Along the way, place the serum and cleansing oil in the storage area above, in either order—before or after closing the drawer."
+    },
+    {
+      "candidate_index": 7,
+      "instruction": "Place the pen in the holder, place the eraser in the open drawer and close it, switch off the lamp, and unplug it. These actions may be performed in any order, except that the lamp must be switched off before it is unplugged.",
+      "task_index": 0
+    },
+    {
+      "candidate_index": 8,
+      "instruction": "Place the wireless mouse, then the wired earphones, in the cardboard box. Cover both items with bubble wrap."
+    },
+    {
+      "candidate_index": 9,
+      "instruction": "Place the pot, with the spatula inside, on the induction cooker so that most of the pot is over the cooking surface. Turn the handle toward the camera, move the spatula to the basket, and cover the pot with the lid."
+    },
+    {
+      "candidate_index": 10,
+      "instruction": "Place the pig bun on the left, the yellow steamed bun on the right, and the black steamed bun at the front of the upper tier of the two already assembled, in that order. Add the empty tier on top, place the large bun in its center, then cover the steamer with the lid."
+    },
+    {
+      "candidate_index": 11,
+      "instruction": "Take the plate, bowl, fork, and knife out of the crate. Place the plate face up in the lower center of the placemat and the bowl open side up in the upper center, in front of the plate. Place the fork on the left and the knife on the right, with the fork tines and knife tip pointing toward the front. Complete these actions in any order."
+    },
+    {
+      "candidate_index": 12,
+      "instruction": "Place the red block on the yellow drawer, the yellow block on the white drawer, and the blue block on the dark-green drawer, in that order."
+    },
+    {
+      "candidate_index": 13,
+      "instruction": "Place the three colored blocks on the plate one at a time in the order red, green, then blue. Tilt the loaded plate to pour all three blocks into the box."
+    }
+  ]
+}

evaluation/real-world/test-30/annotations/tasks.json

@@ -0,0 +1,148 @@
+{
+  "tasks": [
+    {
+      "task_index": 0,
+      "instruction": "Place the pen in the holder, place the eraser in the open drawer and close it, switch off the lamp, and unplug it. These actions may be performed in any order, except that the lamp must be switched off before it is unplugged.",
+      "subtasks": [
+        {
+          "id": "001",
+          "description": "Place the pen in the pen holder."
+        },
+        {
+          "id": "002",
+          "description": "Place the eraser in the storage box."
+        },
+        {
+          "id": "003",
+          "description": "Close the storage box."
+        },
+        {
+          "id": "004",
+          "description": "Turn off the desk lamp."
+        },
+        {
+          "id": "005",
+          "description": "Unplug the desk lamp."
+        }
+      ],
+      "counterfactual_instructions": [
+        {
+          "id": "001",
+          "description": "Place the eraser in the pen holder, place the pen in the open drawer and close it, switch off the lamp, and unplug it. These actions may be performed in any order, except that the lamp must be switched off before it is unplugged."
+        },
+        {
+          "id": "002",
+          "description": "Place the pen in the holder, place the eraser in the open drawer and leave the drawer open, switch off the lamp, and unplug it. These actions may be performed in any order, except that the lamp must be switched off before it is unplugged."
+        },
+        {
+          "id": "003",
+          "description": "Place the pen beside the holder, place the eraser in the open drawer and close it, switch off the lamp, and unplug it. These actions may be performed in any order, except that the lamp must be switched off before it is unplugged."
+        },
+        {
+          "id": "004",
+          "description": "Switch off the lamp and unplug it before placing the pen in the holder or the eraser in the open drawer. Close the drawer after placing the eraser, and switch off the lamp before unplugging it."
+        }
+      ]
+    },
+    {
+      "task_index": 1,
+      "instruction": "Place the pink, white, and yellow flowers consecutively in the vase, in any order. Place the diffuser bottle on the coaster and insert the reed stick into it, in either order. Either diffuser action may occur before or after the flower sequence, but neither may occur between flower placements.",
+      "subtasks": [
+        {
+          "id": "001",
+          "description": "Insert the first flower into the vase."
+        },
+        {
+          "id": "002",
+          "description": "Insert the second flower into the vase."
+        },
+        {
+          "id": "003",
+          "description": "Insert the third flower into the vase."
+        },
+        {
+          "id": "004",
+          "description": "Place the reed diffuser bottle on the coaster."
+        },
+        {
+          "id": "005",
+          "description": "Insert the diffuser reed into the reed diffuser bottle."
+        }
+      ],
+      "counterfactual_instructions": [
+        {
+          "id": "001",
+          "description": "Place the pink, white, and yellow paintbrushes consecutively in the vase, in any order. Place the diffuser bottle on the coaster and insert the reed stick into it, in either order. Either diffuser action may occur before or after the paintbrush sequence, but neither may occur between paintbrush placements."
+        },
+        {
+          "id": "002",
+          "description": "Place the pink, white, and yellow flowers consecutively in the vase, in any order. Place the diffuser bottle on the coaster and leave the reed stick beside the bottle, in either order. Either diffuser action may occur before or after the flower sequence, but neither may occur between flower placements."
+        },
+        {
+          "id": "003",
+          "description": "Place the pink, white, and yellow flowers consecutively in the vase, in any order. Place the diffuser bottle beside the coaster and insert the reed stick into it, in either order. Either diffuser action may occur before or after the flower sequence, but neither may occur between flower placements."
+        },
+        {
+          "id": "004",
+          "description": "Place the pink and white flowers consecutively in the vase, in any order, and leave the yellow flower on the table. Place the diffuser bottle on the coaster and insert the reed stick into it, in either order. Either diffuser action may occur before or after the flower sequence, but neither may occur between flower placements."
+        }
+      ]
+    },
+    {
+      "task_index": 2,
+      "instruction": "Pick up the drumstick and strike the center of the drumhead 5 times continuously.",
+      "subtasks": [
+        {
+          "id": "001",
+          "description": "Strike the center of the drum for the first time."
+        },
+        {
+          "id": "002",
+          "description": "Strike the center of the drum for the second time."
+        },
+        {
+          "id": "003",
+          "description": "Strike the center of the drum for the third time."
+        },
+        {
+          "id": "004",
+          "description": "Strike the center of the drum for the fourth time."
+        },
+        {
+          "id": "005",
+          "description": "Strike the center of the drum for the fifth time."
+        },
+        {
+          "id": "006",
+          "description": "Place the drumstick on the table."
+        }
+      ]
+    },
+    {
+      "task_index": 3,
+      "instruction": "Press the button 5 times, fully releasing it and returning to the waiting position after each press.",
+      "subtasks": [
+        {
+          "id": "001",
+          "description": "Press the button for the first time."
+        },
+        {
+          "id": "002",
+          "description": "Press the button for the second time."
+        },
+        {
+          "id": "003",
+          "description": "Press the button for the third time."
+        },
+        {
+          "id": "004",
+          "description": "Press the button for the fourth time."
+        },
+        {
+          "id": "005",
+          "description": "Press the button for the fifth time."
+        }
+      ]
+    }
+  ]
+}

evaluation/real-world/test-30/annotations/trr_groups.json

@@ -0,0 +1,44 @@
+{
+  "groups": [
+    {
+      "group_index": 0,
+      "task_index": 0,
+      "target_subtask_id": "001",
+      "members": {
+        "ER": 6,
+        "IR": 7,
+        "EC": 8
+      }
+    },
+    {
+      "group_index": 1,
+      "task_index": 0,
+      "target_subtask_id": "002",
+      "members": {
+        "ER": 11,
+        "IR": 10,
+        "EC": 9
+      }
+    },
+    {
+      "group_index": 2,
+      "task_index": 1,
+      "target_subtask_id": "001",
+      "members": {
+        "EC": 13,
+        "ER": 20,
+        "IR": 21
+      }
+    },
+    {
+      "group_index": 3,
+      "task_index": 1,
+      "target_subtask_id": "005",
+      "members": {
+        "IR": 16,
+        "EC": 17,
+        "ER": 23
+      }
+    }
+  ]
+}

evaluation/real-world/test-30/checksums.sha256

@@ -0,0 +1,134 @@
+d41d446fcffb620ea46a39521eaecad93ce1d41bd8a6563b9d2e5892df2af342  README.md
+1d3e85bae54adae57f84573f46b3bb3f9829bb68a9d80b30535441471ff888c4  README_zh.md
+a377ffb50a1adf5354a8d65e9a90175b73cacb4a9800593cf1eaf89885f2f1df  TRAJECTORIES.md
+3cea0680a8709e160e87ec3c01f6b355a3e3956ee68c6b9ef5e628269c792e5a  annotations/csvc_solutions.json
+a3be4d8110975a3a3214b4446a5d31cf41b99a91dc104d5a8f45b3129635806b  annotations/episodes.jsonl
+575007589b74387c0b6204c763f796aef4b0056c0082d8e51e28ba354b44d2bb  annotations/fpl_annotations.json
+c9613c0b52237d7445ec111559bf20c8510537214e9345d2467e5eafefa174d6  annotations/metric_membership.json
+2c6c0d0d41367aecfb516e12467cc0cca086618c7b0aa705abe86651ff6173ce  annotations/task_candidates.json
+34ca39a1e3a17447aa2298f03a87f04d57768169f7803bef910eaf03e2a37b42  annotations/tasks.json
+2ade6283a87d2896d2e5b1cc890033eb9be7e49b070cec45e4a3f48512c146c1  annotations/trr_groups.json
+99852f48ff9b5939df6593c4fcab1ac96ae66b78f46eb28f7629fc8c3ea425bd  data/chunk-000/file-000.parquet
+8a580680014b445b30d62e542fd4a9364284b21282bfbbce527df1da7ad65a10  data/chunk-000/file-001.parquet
+aeb1226a9baaf26f689558e1eeac30a05300aa165d40ea1aa270ef6be1c6b34d  data/chunk-000/file-002.parquet
+c8293c0cc0e5b4f6229b22be1354b07fc4a34738d8bc4d5e8323a19da33aed3d  data/chunk-000/file-003.parquet
+78a2886198b9679612a3631c205fbe3a210b6fdaa4998d089bd1f4a78434eaf2  data/chunk-000/file-004.parquet
+0353d91db5f822884fbf7eff60b52f614c85a258e09a8c44f86ae62636584409  data/chunk-000/file-005.parquet
+49d666174e9db6a30ae22011d90cea3abf12cc90a07695342e58c4164c485986  data/chunk-000/file-006.parquet
+83048c47e5e2a52c3c1cfbde9d910bd58bfe103cfaed34ae3b0dc95b41ea7f78  data/chunk-000/file-007.parquet
+cbc7bb9b2e27591dcc4aea35b03ae04cd29a43ce0ce7993927a1820267d4ce46  data/chunk-000/file-008.parquet
+48c794b345a4634849c57964ac210a15e36e97733b24c8c8c9c72c07fc526013  data/chunk-000/file-009.parquet
+f42764a8e25fe0561bae6125fc9e3c3e46eca0228fb52a57dc8fb70d9ebefbdb  data/chunk-000/file-010.parquet
+f360998a7b2e15de5ba82ac8de37422d12aa0a0c31d9b7638851053fc6bf74af  data/chunk-000/file-011.parquet
+cd75eebfa23f461583b5066008a27c1aac34d832cadace0570595a4743b46fce  data/chunk-000/file-012.parquet
+70c9a888b4deb1f38321f1871fb6ea430af40354ab6136ddca1548cc271e97c3  data/chunk-000/file-013.parquet
+600a6c2c780f920660b54e535a099020ec06d38e62cc3140b71765a8bdfe7c69  data/chunk-000/file-014.parquet
+5d5d7b1d0ae303a613436d7cb6a5d28b63cb8c4eaa93768f68d15e293bccd20e  data/chunk-000/file-015.parquet
+2a9118369cac1474a6551b77e89f2a02a6f0aaa39475cf578e4afde9c26a2a1b  data/chunk-000/file-016.parquet
+c9095e7521265aebc5f0cee88099a6ad836dbbc2c9ee5e265191378b6819be2f  data/chunk-000/file-017.parquet
+5afbc4bad8e1cad07f9de5bfbbfc2543acbeba89907c818d1352153352fa9863  data/chunk-000/file-018.parquet
+f4d1b82f655378af77bb1b1238a43e3a97f6d65515288138363de900c1527068  data/chunk-000/file-019.parquet
+27ff8380398b0e57364a6228c669d841257e292289a2ac93e4028580ca3b694c  data/chunk-000/file-020.parquet
+cd921b15f8695402c749029af243ca8457d38d4db8b2db23b07ce8a09d566569  data/chunk-000/file-021.parquet
+19e3a4ad03f2f5238626a4fa564dc1704d40d74ba74b5a60d82a972802184958  data/chunk-000/file-022.parquet
+0290710ab31511058ce15e0fdd26aadd23ad72e623cedeb9481e63dfff292c62  data/chunk-000/file-023.parquet
+5ece40bfea5a4a2866511e398b2eff5e7607eb8693d06a56e11d38671d3a97d3  data/chunk-000/file-024.parquet
+8b0db6c6d4df6cfd5cc5296616287942d956239eb22391d8ee178f6b86d287f9  data/chunk-000/file-025.parquet
+f49f6499ba121c109d0132f60356434bfdced7ec7487825a9d1663e969db6a06  data/chunk-000/file-026.parquet
+02bf16a1b6a268018a46d7463a6ad00088db771d2cbfc4d3ef93e52d26e1d6d6  data/chunk-000/file-027.parquet
+4805fba1aaf404a4e1449d2837442dc4b66291281556f5eb77856e5573d9ef8b  data/chunk-000/file-028.parquet
+b3700bd29b188b669a13d50fbdac31ce4a0ba9109de07059213851b1df31f05d  data/chunk-000/file-029.parquet
+b623cb5927ecae99f5541c2a9381f697fa2e2600ba567101d579ef2522d019ad  episode_catalog.parquet
+11f1a1564fda431803bf6632907cb1894eceaf456c6ae00ae2839d40ad797a15  meta/episodes/chunk-000/file-000.parquet
+e2da4119e709655b08b2d0102f4af623dcd28d8e43c0cc3f1e2ba2b7cedd855b  meta/info.json
+40fcc7674c74171ffbb93d14b4de1e7b260ec4c9aa5c807d5752dc9b9a558df0  meta/tasks.parquet
+9b3762e91eee2273ad542f1433a1d95d2b076a02d8f97fb1ccf341b9184ae1f5  videos/observation.images.hand_left/chunk-000/file-000.mp4
+68c1b95223f0f95fb665c6e31659e60f50d85e39299c9783fbe80f9ac588cbad  videos/observation.images.hand_left/chunk-000/file-001.mp4
+a3f06152a9766d88b3186b8efda8cfd043ea92bfd730f3208585715f17584aea  videos/observation.images.hand_left/chunk-000/file-002.mp4
+b50168cde7ed7d0a3fd88ba7cca43bf2f673bbfe6959cc3bb11268e52efed4e7  videos/observation.images.hand_left/chunk-000/file-003.mp4
+bf170a9a712ac629d3ddb2af48670071a2af9982c919faa1d434e0da13e76a88  videos/observation.images.hand_left/chunk-000/file-004.mp4
+4cb4f1c92886e1472faebced4b3a011a6db6c2b7db10af0e7939da06f5e92dc7  videos/observation.images.hand_left/chunk-000/file-005.mp4
+6ced2505072a95e7ec0ba53d3c4934c41e927310a72bf621ef0965604356e13b  videos/observation.images.hand_left/chunk-000/file-006.mp4
+4580595fceacf0ca8ec390304b7cd8fbb0ea7004316140f95aa9363841b69191  videos/observation.images.hand_left/chunk-000/file-007.mp4
+ffd83c0205c1c6137f0ce5de7e916b007e47e5ece94cccc43fa3ca27d8cd4692  videos/observation.images.hand_left/chunk-000/file-008.mp4
+54d86e175bfefb977dc9cf60942ba0cf29623abf1ef6f09d1aba904caa9ff30a  videos/observation.images.hand_left/chunk-000/file-009.mp4
+450b27dab7862dcf13d718b08d6ed4830f74ded90710f82b0b42b6f8f384492f  videos/observation.images.hand_left/chunk-000/file-010.mp4
+65665d287bb04c5fda1e161e9b57d2e875f37f6d59bce3b793b080a5fee576b7  videos/observation.images.hand_left/chunk-000/file-011.mp4
+97c64730db5e1b1e220c7d389a9e7c0f10ad5b8c86356fc76ee2c130bfe4e5d9  videos/observation.images.hand_left/chunk-000/file-012.mp4
+163487cfe925cf7df5170e94f434a01b0236e3de61060f75ed9cf11b2e2a3fee  videos/observation.images.hand_left/chunk-000/file-013.mp4
+3fbd761deb01d448d36a30068706d0d9db7a8e8dceb02ef06822d6e87b661d91  videos/observation.images.hand_left/chunk-000/file-014.mp4
+c7e388525836ced3d8f0b211cd2d1cc7c374e79615ee0bf9455d862acf09f489  videos/observation.images.hand_left/chunk-000/file-015.mp4
+a70d6c4a2cdb2c0e7210e47e385261b8bac145fc8f86713f8c2d2c06da191dd8  videos/observation.images.hand_left/chunk-000/file-016.mp4
+9e52b7aaee15ce0f3bcb3b91b5aee4de4a0069f4994669e9491ee2ad8e13a0ce  videos/observation.images.hand_left/chunk-000/file-017.mp4
+cd3a147fa05e33527600b95cc689b9e69a170db88ba1a331294790b1b3ab697a  videos/observation.images.hand_left/chunk-000/file-018.mp4
+bf438bd88b186f8feef3fd159cb9a11108966713b429d1c20eabb47294a08c55  videos/observation.images.hand_left/chunk-000/file-019.mp4
+215b03aeeb7fdaa7b2180de33cb34a71701dc44c298a0fe494915ab1af4dfe57  videos/observation.images.hand_left/chunk-000/file-020.mp4
+afbc72cfdb33b89d93fd7034d2144e4862986c272cd5d75c87b226659d5687af  videos/observation.images.hand_left/chunk-000/file-021.mp4
+14853701a92f93acc3115c408e03368db6e42ed3bacb729a3b09dd2c29a7ff29  videos/observation.images.hand_left/chunk-000/file-022.mp4
+829a68c8ee2ebe1dede77be5d2f49ed0e920eeaec867b92ab9e5c14595b026ea  videos/observation.images.hand_left/chunk-000/file-023.mp4
+a448eae5c911bb7056bc3cb889ea98e224065f2128da467a76b880976aa0473c  videos/observation.images.hand_left/chunk-000/file-024.mp4
+6c811f9fbea810e64ec5c6a839fc6609464a2478369752615d280422cf2f08b8  videos/observation.images.hand_left/chunk-000/file-025.mp4
+36e4d0bac9e55b1e6f990b5560502029877c5ecdb4415aa20884759c60f252a3  videos/observation.images.hand_left/chunk-000/file-026.mp4
+8e58cdd53c57e0108ecee453387e7ab8fadbcbee9cb590b8e62918d595cdbde6  videos/observation.images.hand_left/chunk-000/file-027.mp4
+2d8c052e9b470f78e41861f8802346e620941bad53d55009d4cf4393fa949ea8  videos/observation.images.hand_left/chunk-000/file-028.mp4
+c298463f94f4e5ab8246ca5059c391a7964a1ddbd2debd40941f7ff24f4dbb99  videos/observation.images.hand_left/chunk-000/file-029.mp4
+12a3231b4fa139c360dfde89b3489858a9b663384f73ae4eac202511d0acdf0c  videos/observation.images.hand_right/chunk-000/file-000.mp4
+91e2184699130d419385266a54883f1b5ff54ae72072c7e8eaf993e0fc245ec4  videos/observation.images.hand_right/chunk-000/file-001.mp4
+57f7ccddc7aa4c141f7d2c03ec08b3c2306eead0c68de560a901f048d042f852  videos/observation.images.hand_right/chunk-000/file-002.mp4
+32326520d2858a54296601a5bdc34e47a122f440c4b4ca828d3cb6415aef4e64  videos/observation.images.hand_right/chunk-000/file-003.mp4
+0b12e995ceac0e1296f5365097c56f90d6fd2ecfa2dd6eaa642d67421d42717d  videos/observation.images.hand_right/chunk-000/file-004.mp4
+78e34661cae035a350f76627e3550c21f0e51b0a0bd1c2c75d79e554b010f4cb  videos/observation.images.hand_right/chunk-000/file-005.mp4
+f632d888258647e57f35177e6f05a1feee8b4b2ad063380295bf3b54695e7830  videos/observation.images.hand_right/chunk-000/file-006.mp4
+0d4152f0dc6fe7fc85a24a026b20ce5ca35d1a5dc1934e2a0cd10ca6758e4808  videos/observation.images.hand_right/chunk-000/file-007.mp4
+a7939db311737aafac53a0a2a35da461105c3e16fd13379732d8137cab3e4793  videos/observation.images.hand_right/chunk-000/file-008.mp4
+f2072b7d4b54f27b6511e5878fa171b55ed9080cac7876ef40edde714cc90174  videos/observation.images.hand_right/chunk-000/file-009.mp4
+107deadc9e14aaa6310b3e7c63919ef9e2906b8d2a314ba38043d5f8fdefe7a0  videos/observation.images.hand_right/chunk-000/file-010.mp4
+436f8cb5893064f789d474ca1877a66b78a9a0c03a6195565b21d79ec0f82716  videos/observation.images.hand_right/chunk-000/file-011.mp4
+a93c690664c84d590522adf5dbeddfb6a9da90b00e9ddd1895af10b4312cf6de  videos/observation.images.hand_right/chunk-000/file-012.mp4
+04e0ebc3a9d054e427db4e673618a3ada72b5aa85a9b9a20267c0b06d5309ab1  videos/observation.images.hand_right/chunk-000/file-013.mp4
+23e0659ee3a35f92cae85d934adf45b61085e241a9077b19ec28f7cfc2e8c5a4  videos/observation.images.hand_right/chunk-000/file-014.mp4
+f69fd1bca96654344a76e5d25443086717309c6a0b12ece6ff35b202f0e6e7da  videos/observation.images.hand_right/chunk-000/file-015.mp4
+0eed363bf38f4e0aad74a6bd6973124c61738107bfdc9ce2389d3e3cf00aea3a  videos/observation.images.hand_right/chunk-000/file-016.mp4
+79fd39cab0e2c923fdc721a44c0dff03093fdd868caedd5c3f273715a1c154b3  videos/observation.images.hand_right/chunk-000/file-017.mp4
+318ff11e011aff581461f07189f491220b10ea7781afe402fd0b6e4429ebe72b  videos/observation.images.hand_right/chunk-000/file-018.mp4
+ff023a99e9d9445906258039c4610ecbd51ff5cf530c1bd2c19dae8984d18411  videos/observation.images.hand_right/chunk-000/file-019.mp4
+115d52744c2cb2b118320b351df3783b9b3b2a5dcd18ac61e04d4af04638d5c5  videos/observation.images.hand_right/chunk-000/file-020.mp4
+fcd0560018a4e7aa7fd4b5d8cfa84e1bb3a380b517cbcc97f8aa1084cc4c2013  videos/observation.images.hand_right/chunk-000/file-021.mp4
+47e5af7621c48c51dfe474de0c33e66afcf36e1d11fe9e754e1a69c298b81f84  videos/observation.images.hand_right/chunk-000/file-022.mp4
+583d3c70f098737a68ed2dd5bed6686f2e024c4ec046e15faf68d902091cced7  videos/observation.images.hand_right/chunk-000/file-023.mp4
+5e30834e9805a7a317766f7220b29c37c78856a20011520bbe6e8c25ec3ffcbf  videos/observation.images.hand_right/chunk-000/file-024.mp4
+a092318432e97b36a836d1acbb0dd077515c4221f00d6bd99902a6b1cae8f950  videos/observation.images.hand_right/chunk-000/file-025.mp4
+c6cb26d9efd94592b43934dced513966daa6fb3d2b464724c2447c228a939b66  videos/observation.images.hand_right/chunk-000/file-026.mp4
+2ee62b84acef364807713146426696021cfc0cccba548d0de11c3f1e9a84da12  videos/observation.images.hand_right/chunk-000/file-027.mp4
+074a56c0ee08a9811fd6cada1037f67230be15ad195c80b532b04cbbff593d47  videos/observation.images.hand_right/chunk-000/file-028.mp4
+27a428cc7d0ac36c6056814c259dd6d0c5b8126dec56c34458cc1fffec7bf3f6  videos/observation.images.hand_right/chunk-000/file-029.mp4
+e437f810d1bb65e07682b0199578891dd971c01ac7e02b095b50e41e55b63e30  videos/observation.images.top_head/chunk-000/file-000.mp4
+41e43ea68024a086ae6f54bd9fcc135357d78300d7770981488bfdfedbce71d0  videos/observation.images.top_head/chunk-000/file-001.mp4
+7c932543b4552ee7746f25b11d64b864f4167724c4d3ff4e1f29db55754d62da  videos/observation.images.top_head/chunk-000/file-002.mp4
+56f9969353a0986554bc7d7f21ba3da7c5947ffa0017ce936fe44d31a313aab5  videos/observation.images.top_head/chunk-000/file-003.mp4
+08c9131af51de92f1144fd8dc0d3fcddaf9a9a66df3b95f2eab8ae1eb846631f  videos/observation.images.top_head/chunk-000/file-004.mp4
+01299c60044920ef3a769aae901685cafb0efa9daf3648f4f3501890586d967d  videos/observation.images.top_head/chunk-000/file-005.mp4
+e4513c3e43b1b1b72a90e62c25b6b87139bc8bdbc240fbac763cd6ee2c9c9739  videos/observation.images.top_head/chunk-000/file-006.mp4
+a8e9858679963bb0fa094d508b24210ddfa642962f348249ee518d2dcfaf923d  videos/observation.images.top_head/chunk-000/file-007.mp4
+f015026c87776b62377442c23a457a09ee60cba2687f1310d29216dbfd441b1e  videos/observation.images.top_head/chunk-000/file-008.mp4
+3aa367efa2a5ac352bda1474025154cde696100ee40fc9b795660796afb6176c  videos/observation.images.top_head/chunk-000/file-009.mp4
+a4a56c200d83ed2739fe95be97350bf794b0bf6f6a97ccddf2579cc871153327  videos/observation.images.top_head/chunk-000/file-010.mp4
+2f3597a86e5193e6999ca039a04dedb295f0fd5ce00821809d38a7031622cddf  videos/observation.images.top_head/chunk-000/file-011.mp4
+09dcba122aca52165eab5aa192389557c044d887f40e70307157aa6816bca462  videos/observation.images.top_head/chunk-000/file-012.mp4
+24eab316619c18c65b6e6fb067218dade85eb53e0f6a6585137b68e18bf64cc4  videos/observation.images.top_head/chunk-000/file-013.mp4
+74c48e54c9b39392ea5efa8c729a7dc2f8ae15abaca73e493a4c1cbbaee43c3f  videos/observation.images.top_head/chunk-000/file-014.mp4
+0021049dab38ce60fa5c9629acb28f6cf1c6a786d18e81459e162cdc10c74f63  videos/observation.images.top_head/chunk-000/file-015.mp4
+59cea7ec025b2da258762edda36c1c0a988ac3c1ff2b85e3d4ac1528e0db4ac3  videos/observation.images.top_head/chunk-000/file-016.mp4
+1ebf8926a887444cd6e3934a0cf7ea92b533f970cb21ba7c25e807a256bd0963  videos/observation.images.top_head/chunk-000/file-017.mp4
+037c1dc418b9e9ae8211a7f6455708625a278a525de767ebdd26054f6678b302  videos/observation.images.top_head/chunk-000/file-018.mp4
+3ecd0e12468965b9018ef05e3269eb20b4859fbc9e5066302bd2b9bae327e32d  videos/observation.images.top_head/chunk-000/file-019.mp4
+72998abbcdc0388f87aab591edf9c9f758d37df90d8482f1cce0fdda1d7fe381  videos/observation.images.top_head/chunk-000/file-020.mp4
+556c5370d0cfd0a053d9a5edc6a02f8df3eccc5013b70ac59bb316d2aa8cfa01  videos/observation.images.top_head/chunk-000/file-021.mp4
+d860b43b14980ecbcd954f8c4b5a1ae5c0367e11f16328769d5af81b50daddb6  videos/observation.images.top_head/chunk-000/file-022.mp4
+605577d18a5b99f4b65c372972ece82f5a9f21de5b6c65906cc19dee21a4b818  videos/observation.images.top_head/chunk-000/file-023.mp4
+45da308115ea0d93131e36d6ca6572a3d2d7bbf3cf04714383748e466fb1c4d7  videos/observation.images.top_head/chunk-000/file-024.mp4
+05e5c14a64c4fda905697ad92988e3259e09c93ccb34fe02efa730225c82098a  videos/observation.images.top_head/chunk-000/file-025.mp4
+05caad12603d32d25a241ce78bc4a7a07570fdcb10ba71f9a2003322d7f4dcdc  videos/observation.images.top_head/chunk-000/file-026.mp4
+288086a93fdfc61270f46f1d090d4f303f8d64362ed1a3614bf7e6926bdc301f  videos/observation.images.top_head/chunk-000/file-027.mp4
+5cdb8b75e85d5742d67f06aaa02ca5f0809efebb45a3877ea3cba1d90301a570  videos/observation.images.top_head/chunk-000/file-028.mp4
+5a4f4df8ed2a47812773d4c31185ea33a9b8b408d454264d4409ec7079fb4a1a  videos/observation.images.top_head/chunk-000/file-029.mp4

evaluation/real-world/test-30/data/chunk-000/file-000.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99852f48ff9b5939df6593c4fcab1ac96ae66b78f46eb28f7629fc8c3ea425bd
+size 11976

evaluation/real-world/test-30/data/chunk-000/file-001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a580680014b445b30d62e542fd4a9364284b21282bfbbce527df1da7ad65a10
+size 11768

evaluation/real-world/test-30/data/chunk-000/file-002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb1226a9baaf26f689558e1eeac30a05300aa165d40ea1aa270ef6be1c6b34d
+size 10923

evaluation/real-world/test-30/data/chunk-000/file-003.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8293c0cc0e5b4f6229b22be1354b07fc4a34738d8bc4d5e8323a19da33aed3d
+size 9962

evaluation/real-world/test-30/data/chunk-000/file-004.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78a2886198b9679612a3631c205fbe3a210b6fdaa4998d089bd1f4a78434eaf2
+size 9333

evaluation/real-world/test-30/data/chunk-000/file-005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0353d91db5f822884fbf7eff60b52f614c85a258e09a8c44f86ae62636584409
+size 10769

evaluation/real-world/test-30/data/chunk-000/file-006.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49d666174e9db6a30ae22011d90cea3abf12cc90a07695342e58c4164c485986
+size 12730

evaluation/real-world/test-30/data/chunk-000/file-007.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83048c47e5e2a52c3c1cfbde9d910bd58bfe103cfaed34ae3b0dc95b41ea7f78
+size 12876

evaluation/real-world/test-30/data/chunk-000/file-008.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbc7bb9b2e27591dcc4aea35b03ae04cd29a43ce0ce7993927a1820267d4ce46
+size 10089

evaluation/real-world/test-30/data/chunk-000/file-009.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48c794b345a4634849c57964ac210a15e36e97733b24c8c8c9c72c07fc526013
+size 9327

evaluation/real-world/test-30/data/chunk-000/file-010.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f42764a8e25fe0561bae6125fc9e3c3e46eca0228fb52a57dc8fb70d9ebefbdb
+size 10365

evaluation/real-world/test-30/data/chunk-000/file-011.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f360998a7b2e15de5ba82ac8de37422d12aa0a0c31d9b7638851053fc6bf74af
+size 10231

evaluation/real-world/test-30/data/chunk-000/file-012.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd75eebfa23f461583b5066008a27c1aac34d832cadace0570595a4743b46fce
+size 10323

evaluation/real-world/test-30/data/chunk-000/file-013.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70c9a888b4deb1f38321f1871fb6ea430af40354ab6136ddca1548cc271e97c3
+size 10082

evaluation/real-world/test-30/data/chunk-000/file-014.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:600a6c2c780f920660b54e535a099020ec06d38e62cc3140b71765a8bdfe7c69
+size 10378

evaluation/real-world/test-30/data/chunk-000/file-015.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d5d7b1d0ae303a613436d7cb6a5d28b63cb8c4eaa93768f68d15e293bccd20e
+size 11784

evaluation/real-world/test-30/data/chunk-000/file-016.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a9118369cac1474a6551b77e89f2a02a6f0aaa39475cf578e4afde9c26a2a1b
+size 12126

evaluation/real-world/test-30/data/chunk-000/file-017.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9095e7521265aebc5f0cee88099a6ad836dbbc2c9ee5e265191378b6819be2f
+size 9638

evaluation/real-world/test-30/data/chunk-000/file-018.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5afbc4bad8e1cad07f9de5bfbbfc2543acbeba89907c818d1352153352fa9863
+size 11628

evaluation/real-world/test-30/data/chunk-000/file-019.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4d1b82f655378af77bb1b1238a43e3a97f6d65515288138363de900c1527068
+size 10329

evaluation/real-world/test-30/data/chunk-000/file-020.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27ff8380398b0e57364a6228c669d841257e292289a2ac93e4028580ca3b694c
+size 13839

evaluation/real-world/test-30/data/chunk-000/file-021.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd921b15f8695402c749029af243ca8457d38d4db8b2db23b07ce8a09d566569
+size 12585

evaluation/real-world/test-30/data/chunk-000/file-022.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19e3a4ad03f2f5238626a4fa564dc1704d40d74ba74b5a60d82a972802184958
+size 10972

evaluation/real-world/test-30/data/chunk-000/file-023.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0290710ab31511058ce15e0fdd26aadd23ad72e623cedeb9481e63dfff292c62
+size 14078

evaluation/real-world/test-30/data/chunk-000/file-024.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ece40bfea5a4a2866511e398b2eff5e7607eb8693d06a56e11d38671d3a97d3
+size 7157

evaluation/real-world/test-30/data/chunk-000/file-025.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b0db6c6d4df6cfd5cc5296616287942d956239eb22391d8ee178f6b86d287f9
+size 7020

evaluation/real-world/test-30/data/chunk-000/file-026.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f49f6499ba121c109d0132f60356434bfdced7ec7487825a9d1663e969db6a06
+size 7290

evaluation/real-world/test-30/data/chunk-000/file-027.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02bf16a1b6a268018a46d7463a6ad00088db771d2cbfc4d3ef93e52d26e1d6d6
+size 7142

evaluation/real-world/test-30/data/chunk-000/file-028.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4805fba1aaf404a4e1449d2837442dc4b66291281556f5eb77856e5573d9ef8b
+size 7276

evaluation/real-world/test-30/data/chunk-000/file-029.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3700bd29b188b669a13d50fbdac31ce4a0ba9109de07059213851b1df31f05d
+size 6865

evaluation/real-world/test-30/episode_catalog.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b623cb5927ecae99f5541c2a9381f697fa2e2600ba567101d579ef2522d019ad
+size 6191

evaluation/real-world/test-30/meta/episodes/chunk-000/file-000.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11f1a1564fda431803bf6632907cb1894eceaf456c6ae00ae2839d40ad797a15
+size 10451

evaluation/real-world/test-30/meta/info.json

@@ -0,0 +1,122 @@
+{
+  "codebase_version": "v3.0",
+  "robot_type": null,
+  "total_episodes": 30,
+  "total_frames": 27498,
+  "total_tasks": 4,
+  "chunks_size": 1000,
+  "data_files_size_in_mb": 100,
+  "video_files_size_in_mb": 200,
+  "fps": 30,
+  "splits": {
+    "train": "0:30"
+  },
+  "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
+  "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
+  "features": {
+    "timestamp": {
+      "dtype": "float32",
+      "shape": [
+        1
+      ],
+      "names": null
+    },
+    "frame_index": {
+      "dtype": "int64",
+      "shape": [
+        1
+      ],
+      "names": null
+    },
+    "episode_index": {
+      "dtype": "int64",
+      "shape": [
+        1
+      ],
+      "names": null
+    },
+    "index": {
+      "dtype": "int64",
+      "shape": [
+        1
+      ],
+      "names": null
+    },
+    "task_index": {
+      "dtype": "int64",
+      "shape": [
+        1
+      ],
+      "names": null
+    },
+    "observation.images.top_head": {
+      "dtype": "video",
+      "shape": [
+        400,
+        640,
+        3
+      ],
+      "names": [
+        "height",
+        "width",
+        "channel"
+      ],
+      "info": {
+        "video.height": 400,
+        "video.width": 640,
+        "video.codec": "hevc",
+        "video.pix_fmt": "yuv420p",
+        "video.is_depth_map": false,
+        "video.fps": 30,
+        "video.channels": 3,
+        "has_audio": false
+      }
+    },
+    "observation.images.hand_left": {
+      "dtype": "video",
+      "shape": [
+        528,
+        640,
+        3
+      ],
+      "names": [
+        "height",
+        "width",
+        "channel"
+      ],
+      "info": {
+        "video.height": 528,
+        "video.width": 640,
+        "video.codec": "hevc",
+        "video.pix_fmt": "yuv420p",
+        "video.is_depth_map": false,
+        "video.fps": 30,
+        "video.channels": 3,
+        "has_audio": false
+      }
+    },
+    "observation.images.hand_right": {
+      "dtype": "video",
+      "shape": [
+        528,
+        640,
+        3
+      ],
+      "names": [
+        "height",
+        "width",
+        "channel"
+      ],
+      "info": {
+        "video.height": 528,
+        "video.width": 640,
+        "video.codec": "hevc",
+        "video.pix_fmt": "yuv420p",
+        "video.is_depth_map": false,
+        "video.fps": 30,
+        "video.channels": 3,
+        "has_audio": false
+      }
+    }
+  }
+}

evaluation/real-world/test-30/meta/tasks.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40fcc7674c74171ffbb93d14b4de1e7b260ec4c9aa5c807d5752dc9b9a558df0
+size 2530

evaluation/real-world/test-30/videos/observation.images.hand_left/chunk-000/file-000.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b3762e91eee2273ad542f1433a1d95d2b076a02d8f97fb1ccf341b9184ae1f5
+size 7277281

evaluation/real-world/test-30/videos/observation.images.hand_left/chunk-000/file-001.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68c1b95223f0f95fb665c6e31659e60f50d85e39299c9783fbe80f9ac588cbad
+size 7626913

evaluation/real-world/test-30/videos/observation.images.hand_left/chunk-000/file-002.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3f06152a9766d88b3186b8efda8cfd043ea92bfd730f3208585715f17584aea
+size 6903974