The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 84, in _split_generators
raise ValueError(
...<2 lines>...
)
ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
egoR1 — 完整目录归档 / Full Directory Archive
本仓库以 TAR 分卷原样分发 egoR1_dataset/,不是 Parquet 转换版。下载、拼接、解压后即可恢复原目录结构、图片、训练和评测 JSONL、配置、代码、测试、报告及离线基线结果。
This repository distributes the complete egoR1_dataset/ directory as consecutive parts of one uncompressed TAR archive. It is designed for exact directory restoration, not automatic Hugging Face Dataset Viewer loading.
数据 / Data
- 任务:输入 8 张历史帧、1 张当前帧和 5 个候选,预测下一次手首次接触对象及其当前帧位置;Temporal 协议额外预测 TTC。
- v1.0:36,410 个样本,train / validation / test 为 29,106 / 3,670 / 3,634。
- 数据集的许可、限制、字段与完整用法见解压后的
egoR1_dataset/README.md和DATASET_CARD.md。 - 图片原始字节不重新压缩,所有数据相对路径不变。所有图片均为归档内实体文件,不依赖旧源目录。
- 包含测试标签和未来帧,是开放标签研究发布,不是隐藏测试服务。评测模型只能消费 inputs.jsonl 中允许的内容。
文件 / Files
archives/v1.0/egoR1_dataset.tar.part-000
archives/v1.0/egoR1_dataset.tar.part-001
...
archive_info.json # 分卷顺序、大小及完整 TAR SHA256
SHA256SUMS # 分卷 SHA256
archive_files.sha256 # 解压后全部文件 SHA256,含原 checksums.sha256
source_snapshot.json # 打包源目录快照统计
verification_report.json # 完整归档验证结果
restore.sh # 下载后恢复并逐文件校验
每个分卷最大 4 GiB,最后一个通常不足 4 GiB。分卷是同一个 TAR 的连续字节,不能分别解压。准确文件数与体积以 archive_info.json 为准。
下载 / Download
将 <HF_REPO_ID> 替换为实际仓库名,例如 your-org/egoR1:
hf download <HF_REPO_ID> --repo-type dataset --local-dir ./egoR1_download
如发布者设置了版本 tag,可增加 --revision v1.0。私有或 gated 仓库须先按仓库要求登录并获取访问权限。
一键恢复 / Restore
需要 Linux/macOS 的 bash、tar、sha256sum(macOS 需安装对应工具)和 Python 3。无须安装模型训练依赖。
bash ./egoR1_download/restore.sh ./restored
输出目录:
restored/egoR1_dataset/
├── assets/
├── configs/
├── schemas/
├── scripts/
├── tests/
├── releases/
├── runs/
├── README.md
├── DATASET_CARD.md
├── requirements.txt
└── checksums.sha256
脚本先核验分卷,流式解压,再逐文件 SHA256 校验。目标中已有 egoR1_dataset/ 时拒绝覆盖;失败后的不完整目录须自行检查,再选择新的目标目录重试。
下载和解压内容合计约需 160 GB 可用空间(以实际归档及内容大小为准,建议留足余量)。流式拼接无需额外生成约 79 GB 的中间 TAR。
手动恢复
cd egoR1_download
sha256sum -c SHA256SUMS
mkdir -p ../restored
# 确认 ../restored/egoR1_dataset 不存在,避免覆盖其他文件。
cat archives/v1.0/egoR1_dataset.tar.part-* | tar -xf - -C ../restored
(cd ../restored && sha256sum -c ../egoR1_download/archive_files.sha256 --quiet)
(cd ../restored/egoR1_dataset && sha256sum -c checksums.sha256 --quiet)
验证整个拼接 TAR(可选,预期值在 archive_info.json):
cat archives/v1.0/egoR1_dataset.tar.part-* | sha256sum
一致性保证 / Integrity
发布验证重新读取所有分卷,流式重建 TAR 并核对每个普通文件的 SHA256、路径、大小、类型、权限和修改时间,也验证了目录条目、各分卷与整个 TAR 的哈希。此方法不另外落盘一份约 79 GB 数据;验证结论见 verification_report.json。
目录与文件内容原样保留;恢复后的 UID/GID 属于本机用户,不要求与打包机器一致。纳秒时间戳的实际恢复精度取决于 tar 与文件系统。分卷校验成功后不要在解压期间修改下载文件。
发布说明 / Publication notes
本地打包不等于已对外授权。上传者应在公开发布前确认源图片、标注及教师生成产物的分发权利,并在 HF 仓库补充经确认的许可和引用信息;本归档不自行授予新的许可。
上传整个本发布目录即可,无需再次上传源目录里的几十万张散图。保留这些分卷文件名和目录结构,不改写冻结归档。后续数据更新请发布新的版本路径和 tag。
- Downloads last month
- 48