The dataset viewer is not available for this dataset.
Error code: JobManagerCrashedError
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.
Mirror of mikause/tsinghua_topic_dataset from ModelScope. All credit to the original authors. Not an official release from them. License unchanged (apache-2.0).
1769 file(s) listed on ModelScope could not be mirrored because ModelScope serves them as empty (0 bytes), e.g.
agibot_alpha_joint_v1_overlay/payload/data/chunk-001/file-536.parquet.
Layout
Hugging Face allows at most 10,000 files per folder, so the folders below were reorganised for this mirror. File contents are unchanged.
EgoVerse-official-20260922/controls/69/(80,866 files): split intoEgoVerse-official-20260922/controls/69/part-0000/...EgoVerse-official-20260922/controls/69/part-0008/(up to 9,000 files each, in original name order).EgoVerse-official-20260922/videos/69/(80,866 files): split intoEgoVerse-official-20260922/videos/69/part-0000/...EgoVerse-official-20260922/videos/69/part-0008/(up to 9,000 files each, in original name order).EgoVerse-official-20260922/pairs/69/(40,433 small files): packed intoEgoVerse-official-20260922/pairs/69/shard-00000.tar(1 tar shard); members keep their original paths.EgoVerse-official-20260922/video_validation/69/(13,437 small files): packed intoEgoVerse-official-20260922/video_validation/69/shard-00000.tar(1 tar shard); members keep their original paths.
manifest.jsonl maps every moved file's original ModelScope path to its new location (new_path, or tar + member), with its size. All other files keep their original paths.
Read a tar shard: tarfile.open(hf_hub_download("ml-resources/tsinghua_topic_dataset", "EgoVerse-official-20260922/pairs/69/shard-00000.tar", repo_type="dataset")).extractall("out") (or stream it with webdataset.WebDataset(url)).
For dataset file metadata and data files, please visit the "Files and versions" tab.
The current dataset card uses the default template, and the dataset contributor has not provided a more detailed dataset introduction. However, you can download the dataset using the following GIT Clone command or through the ModelScope SDK.
DOWNLOAD METHOD
:modelscope-code[]{type="sdk"} :modelscope-code[]{type="git"}
- Downloads last month
- 223