FreDERicia commited on
Commit
68c20ff
·
verified ·
1 Parent(s): 7c71398

Use English README with SWAM Pretrain styling

Browse files
Files changed (1) hide show
  1. README.md +24 -16
README.md CHANGED
@@ -13,33 +13,37 @@ tags:
13
 
14
  # SWAM-Posttrain
15
 
16
- [![SWAM 项目封面](https://huggingface.co/datasets/FreDERicia/SWAM-Pretrain/resolve/main/visual-previews/swam-cover-cropped.png)](https://char-lotte-xia.github.io/swam-official/)
17
 
18
- [项目主页](https://char-lotte-xia.github.io/swam-official/) · [论文](https://arxiv.org/abs/2611.12870) · [GitHub](https://github.com/shaaaaron/SWAM)
19
 
20
- 本仓库提供 **SWAM — Scaling World Action Models with Cross-Modal Alignment** 使用的 RoboTwin Clean 后训练数据。
21
 
22
- **共 50 个任务、2,500 条轨迹,约 3.94 GB,按任务打包为 50 个 TAR 归档。** [浏览全部数据](https://huggingface.co/datasets/FreDERicia/SWAM-Posttrain/tree/main/data)
23
 
24
- ## 数据内容
25
 
26
- 每个任务保留 LeRobot 风格的 `data/`、`videos/` 和 `meta/` 目录,包含视频、语言指令和统计信息。数据保留原始轨迹编号,**不包含 latents**。
27
 
28
- 原始数据中有 **16 条轨迹需要排除**,记录在各任务的 `meta/episode_exclusions_20261023.jsonl` 中。它们仍保留在归档内;训练时应用排除清单后,共 **2,484 条可用轨迹**。语言指令使用 `meta/tasks_v1.jsonl`。
29
 
30
- ## 下载与解压
31
 
32
- 安装 Hugging Face CLI。仓库当前为 private,下载前需使用具备访问权限的账号登录。
33
 
34
  ```bash
35
  pip install -U huggingface_hub
36
  hf auth login
 
 
 
37
 
 
38
  hf download FreDERicia/SWAM-Posttrain \
39
  --repo-type dataset --local-dir ./SWAM-Posttrain
40
  ```
41
 
42
- 在 Linux 或 macOS 上,将所有任务归档解压到同一个目录:
43
 
44
  ```bash
45
  mkdir -p ./lerobot_robotwin_eef_clean_50
@@ -48,17 +52,21 @@ for archive in ./SWAM-Posttrain/data/*.tar; do
48
  done
49
  ```
50
 
51
- 解压后的结构为 `lerobot_robotwin_eef_clean_50/<任务目录>/{data,videos,meta}/`。
 
 
 
 
52
 
53
- ## 许可与配套资源
54
 
55
- 原始数据声明采用 [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/);使用时请遵守上游许可与署名要求。原始说明保留在 [SOURCE_README.md](https://huggingface.co/datasets/FreDERicia/SWAM-Posttrain/blob/main/SOURCE_README.md)。
56
 
57
- [模型权重](https://huggingface.co/FreDERicia/SWAM) · [预训练数据](https://huggingface.co/datasets/FreDERicia/SWAM-Pretrain)
58
 
59
- ## 引用
60
 
61
- 如果本数据集对你的研究有帮助,请引用 SWAM 论文。
62
 
63
  ```bibtex
64
  @misc{zhou2026swam,
 
13
 
14
  # SWAM-Posttrain
15
 
16
+ [![SWAM project cover](https://huggingface.co/datasets/FreDERicia/SWAM-Pretrain/resolve/main/visual-previews/swam-cover-cropped.png)](https://char-lotte-xia.github.io/swam-official/)
17
 
18
+ <div style="display: flex; justify-content: center; align-items: center; gap: 8px; flex-wrap: wrap; margin: 12px 0 20px;"><a style="display: inline-flex; margin: 0; line-height: 0;" href="https://github.com/shaaaaron/SWAM"><img src="https://huggingface.co/datasets/FreDERicia/SWAM-Pretrain/resolve/main/visual-previews/link-github.png" alt="GitHub code" width="108" height="35" style="display: inline-block; margin: 0; vertical-align: middle;"></a><a style="display: inline-flex; margin: 0; line-height: 0;" href="https://char-lotte-xia.github.io/swam-official/"><img src="https://huggingface.co/datasets/FreDERicia/SWAM-Pretrain/resolve/main/visual-previews/link-website.png" alt="Project website" width="108" height="35" style="display: inline-block; margin: 0; vertical-align: middle;"></a><a style="display: inline-flex; margin: 0; line-height: 0;" href="https://arxiv.org/abs/2611.12870"><img src="https://huggingface.co/datasets/FreDERicia/SWAM-Pretrain/resolve/main/visual-previews/link-arxiv.png" alt="arXiv paper" width="108" height="35" style="display: inline-block; margin: 0; vertical-align: middle;"></a></div>
19
 
20
+ This repository provides the RoboTwin Clean posttraining data used in SWAM: Scaling World Action Models with Cross-Modal Alignment.
21
 
22
+ The release contains 50 tasks and 2,500 episodes, distributed as 50 TAR archives totaling approximately 3.94 GB. [Browse all data](https://huggingface.co/datasets/FreDERicia/SWAM-Posttrain/tree/main/data)
23
 
24
+ ## Data format and directory structure
25
 
26
+ Each task follows a LeRobot-style layout with `data/`, `videos/`, and `meta/` directories, containing actions and states, videos, language instructions, and statistics. Original episode IDs are preserved. Latents are not included.
27
 
28
+ The archives retain 16 episodes listed in the per-task `meta/episode_exclusions_20261023.jsonl` files. Apply these exclusions during training to obtain 2,484 eligible episodes. Task instructions are stored in `meta/tasks_v1.jsonl`.
29
 
30
+ ## Download and extract
31
 
32
+ Install the Hugging Face CLI and sign in with an account that has access to this private repository.
33
 
34
  ```bash
35
  pip install -U huggingface_hub
36
  hf auth login
37
+ ```
38
+
39
+ Download the dataset.
40
 
41
+ ```bash
42
  hf download FreDERicia/SWAM-Posttrain \
43
  --repo-type dataset --local-dir ./SWAM-Posttrain
44
  ```
45
 
46
+ Extract all task archives into the same directory on Linux or macOS.
47
 
48
  ```bash
49
  mkdir -p ./lerobot_robotwin_eef_clean_50
 
52
  done
53
  ```
54
 
55
+ The extracted layout is `lerobot_robotwin_eef_clean_50/<task-directory>/{data,videos,meta}/`.
56
+
57
+ ## Source terms and attribution
58
+
59
+ The source data declares a [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. Please follow the upstream license and attribution requirements. The original documentation is retained in [SOURCE_README.md](https://huggingface.co/datasets/FreDERicia/SWAM-Posttrain/blob/main/SOURCE_README.md).
60
 
61
+ ## Related resources
62
 
63
+ [Model checkpoints](https://huggingface.co/FreDERicia/SWAM) · [Pretraining data](https://huggingface.co/datasets/FreDERicia/SWAM-Pretrain)
64
 
65
+ ## Code and citation
66
 
67
+ Code is available in the [SWAM code repository](https://github.com/shaaaaron/SWAM).
68
 
69
+ If you use this dataset, please cite the SWAM paper.
70
 
71
  ```bibtex
72
  @misc{zhou2026swam,