Hugging Face Datasets connector
Set up the Hugging Face Datasets connector in Kaivo: authentication, configuration, the 2 BigQuery tables it syncs, and answers to common questions.
Hugging Face Datasets connector pricing, tier and credits on kaivo.ioKaivo is a fully managed data platform that syncs your Hugging Face Datasets data into a Google BigQuery warehouse and keeps it up to date automatically. There is no pipeline to build and no infrastructure to run, so you can spend your time analysing your data from Hugging Face Datasets instead of moving it.
What is the Hugging Face Datasets connector
Section titled “What is the Hugging Face Datasets connector”Sync your Hugging Face dataset rows into BigQuery with Kaivo to analyse dataset content alongside the rest of your data.
| Category | Files & Databases |
|---|---|
| Authentication | Other |
| Setup | Self-service |
Getting started with the Hugging Face Datasets connector
Section titled “Getting started with the Hugging Face Datasets connector”- Sign up for Kaivo and create a workspace.
- Connect your Hugging Face Datasets account.
- Choose which tables to sync.
- Wait for the initial sync to finish.
- Query your data in BigQuery or your favourite AI or BI tool.
Authenticating Hugging Face Datasets
Section titled “Authenticating Hugging Face Datasets”Follow the steps below to connect your Hugging Face Datasets account.
Configuring the Hugging Face Datasets connector
Section titled “Configuring the Hugging Face Datasets connector”When you set up the connector, you provide:
| Field | Description |
|---|---|
| Dataset Name | |
| Dataset Subsets | |
| Dataset Subsets to import. Will import all of them if nothing is provided (see https://huggingface.co/docs/dataset-viewer/en/configs_and_splits for more details) |
| | Dataset Splits |
Splits to import. Will import all of them if nothing is provided (see https://huggingface.co/docs/dataset-viewer/en/configs_and_splits for more details)
|
Tables and columns synced from Hugging Face Datasets
Section titled “Tables and columns synced from Hugging Face Datasets”Kaivo syncs 2 tables from Hugging Face Datasets into a dedicated dataset in your BigQuery warehouse. Click any table to see its columns and types.
rows (5 columns)
| Column | Type | Description |
|---|---|---|
_kaivo_id |
STRING |
Primary key that uniquely identifies the row. Auto-generated by Kaivo. |
config |
STRING |
|
row__tokens |
FLOAT64 |
|
row_idx |
FLOAT64 |
|
_kaivo_extracted_at |
TIMESTAMP |
Timestamp that shows when the row was extracted. Auto-generated by Kaivo. |
Subtable: rows__row__messages
Section titled “Subtable: rows__row__messages”| Column | Type | Description |
|---|---|---|
_kaivo_parent_id |
STRING |
Foreign key referencing _kaivo_id in the parent table. Auto-generated by Kaivo. |
_kaivo_id |
STRING |
Primary key that uniquely identifies the row. Auto-generated by Kaivo. |
content |
STRING |
|
role |
STRING |
|
_kaivo_extracted_at |
TIMESTAMP |
Timestamp that shows when the row was extracted. Auto-generated by Kaivo. |
Subtable: rows__truncated_cells
Section titled “Subtable: rows__truncated_cells”| Column | Type | Description |
|---|---|---|
_kaivo_parent_id |
STRING |
Foreign key referencing _kaivo_id in the parent table. Auto-generated by Kaivo. |
_kaivo_id |
STRING |
Primary key that uniquely identifies the row. Auto-generated by Kaivo. |
value |
JSON |
|
_kaivo_extracted_at |
TIMESTAMP |
Timestamp that shows when the row was extracted. Auto-generated by Kaivo. |
splits (6 columns)
| Column | Type | Description |
|---|---|---|
_kaivo_id |
STRING |
Primary key that uniquely identifies the row. Auto-generated by Kaivo. |
config |
STRING |
|
dataset |
STRING |
|
key |
STRING |
|
split |
STRING |
|
_kaivo_extracted_at |
TIMESTAMP |
Timestamp that shows when the row was extracted. Auto-generated by Kaivo. |
How the Hugging Face Datasets sync works
Section titled “How the Hugging Face Datasets sync works”After the first load, Kaivo keeps your BigQuery warehouse up to date for you. Where Hugging Face Datasets supports it, each sync pulls only new and changed records so it stays fast; otherwise it refreshes the whole table. Every record keeps its original ID, so you won’t get duplicate rows.
Frequently asked questions
Section titled “Frequently asked questions”How long does the initial sync take for Hugging Face Datasets?
Section titled “How long does the initial sync take for Hugging Face Datasets?”It depends on how much history is in your Hugging Face Datasets account. Most initial syncs finish within minutes, while large accounts can take a few hours. After that, syncs only fetch new and changed records, so they’re much faster.
Can I sync only some tables or columns?
Section titled “Can I sync only some tables or columns?”Yes. You pick which tables to sync when you set up the connection and can change the selection later. Tables you don’t select are never copied to your warehouse.
What happens when Hugging Face Datasets’s schema changes?
Section titled “What happens when Hugging Face Datasets’s schema changes?”New fields are never added automatically. You choose which fields to sync, so data you haven’t selected (sensitive personal data, for example) never lands in your warehouse. When a new field appears, it becomes available for you to add. What happens to removed or renamed fields depends on a table’s sync mode: full-refresh tables always match what’s currently in Hugging Face Datasets, so dropped fields disappear, while incremental tables keep their existing columns and history, so an old field stays and newly added fields fill in over time.
How do I handle GDPR or data deletion requests?
Section titled “How do I handle GDPR or data deletion requests?”Your data lives in your own Kaivo-managed BigQuery warehouse, so the most direct option is to delete or anonymise specific records right in BigQuery. If you delete data in Hugging Face Datasets instead, full-refresh tables drop it on the next sync, while incremental tables keep it, so you would remove the row in BigQuery or ask us to run a full refresh. To remove everything, delete the Hugging Face Datasets connector in Kaivo and all of its synced data is deleted with it.
Common use cases for Hugging Face Datasets data
Section titled “Common use cases for Hugging Face Datasets data”Dataset analysis
Section titled “Dataset analysis”Use rows to explore and analyse dataset content in BigQuery.
Split comparison
Section titled “Split comparison”Use splits to compare train, validation, and test sets.
Combine with other data
Section titled “Combine with other data”Join dataset rows with your own data for custom analysis.
Use Hugging Face Datasets data in your AI and BI tools
Section titled “Use Hugging Face Datasets data in your AI and BI tools”Once Hugging Face Datasets data lands in your Kaivo-managed BigQuery warehouse, you can explore it with AI tools or any BI tool that connects to BigQuery. Here’s how the most common destinations work with Hugging Face Datasets data.
Claude
Section titled “Claude”Use Kaivo’s MCP server to give Claude secure, workspace-scoped access to your data. Setup guide →
Power BI
Section titled “Power BI”Microsoft’s BI tool with a native BigQuery connector. Supports direct query and scheduled refresh. Setup guide →
Data Studio
Section titled “Data Studio”Free Google BI tool with native BigQuery support. One-click connection to your Kaivo warehouse; great for SMB teams on Google Workspace. Setup guide →
Tableau
Section titled “Tableau”The premium analytics standard, with native BigQuery integration. Setup guide →
Google Sheets
Section titled “Google Sheets”Use Connected Sheets to query BigQuery directly from a spreadsheet, with no SQL. Setup guide →
Connect via Power Query’s BigQuery connector. Setup guide →
Metabase
Section titled “Metabase”Open-source BI tool with strong BigQuery support. Setup guide →
See our pricing page for Hugging Face Datasets connector pricing and plan details.
Related connectors
Section titled “Related connectors”- Google Sheets: Sync Google Sheets to BigQuery.
- M-Files: Sync M-Files to BigQuery.
- Microsoft SQL Server: Sync Microsoft SQL Server to BigQuery.
- MySQL: Sync MySQL to BigQuery.
- Postgres: Sync Postgres to BigQuery.
- PxWeb: Sync PxWeb to BigQuery.