Skip to content

Browse Datasets (Dataset hub) ​

Find datasets provided by ThakiCloud and datasets users uploaded, check details and data previews, and download them.

Step 1. Open Dataset hub ​

In the left sidebar, click Datasets → Dataset hub. Follow the numbers on the screens below.

SNB ​

Hub menus. Switch between models and datasets.

SNB
No.FeatureItemDescription
1-1ModelsModel hub· Click to go to the model list.
1-2DatasetsDataset hub· Current screen. Browse datasets in Hub as cards.

Search · Filters ​

Narrow the dataset list by source, type, and name.

Search · Filters
No.FeatureItemDescription
2-1Source tabsAll · Thaki · Custom· Thaki shows datasets provided by ThakiCloud; Custom shows datasets users uploaded. Only Custom datasets can be edited or deleted.
2-2Type chipsAll · Text generation · Tabular· Filter by dataset type.
2-3SearchFind datasets· Search by dataset name.

Upload ​

Click to open the upload drawer on the right.

Upload

Dataset card ​

Each dataset is one card.

Dataset card
No.FeatureItemDescription
4-1Card infoName · Description · Tags · Downloads · Language· Dataset name and description, scope (Global, Project) and size tags, download count, and language.
4-2View detailsⓘ button· Click to go to the dataset detail (Data viewer) screen.
4-3DownloadDownload· Download the dataset file.

Cards on the Custom tab also have Edit and Delete buttons. See Edit a dataset and Delete a dataset.


Step 2. View dataset details ​

Click the ⓘ button on a card (number 4-2) to go to the detail screen (Data viewer).

Summary ​

Top of the detail screen.

Summary
No.FeatureItemDescription
1-1NameDataset name· Name of the dataset you are viewing.
1-2DownloadDownload· Download the dataset file.

Details tab ​

The Details tab shows basic information; the Data tab previews the data. Formats that cannot be previewed show No data available.

Details tab
No.FeatureItemDescription
2-1TabsDetails · Data· Switch between details and data preview.
2-2Basic informationFilesFiles in the dataset.
Supported methodsTraining methods chosen at upload (SFT, DPO, etc.).
LicenseLicense.
Data splits · FeaturesData splits (train/test, etc.) and column information.
Source URLOriginal source address.
Created at · Updated atCreation time and last update time.

Data tab — preview by data format ​

FormatPreview
TabularFile name and a table, 10 rows per page
TextFile name and up to 100 lines
ConversationFile name and statistics (average messages per sample, average content length, messages per role)

Step 3. Download a dataset ​

Click Download on a card or on the detail screen. Both Thaki and Custom datasets can be downloaded.

The New Standard for AI Infrastructure