Gena means knowledge, information, or data in Tulu. It is TuluAI’s data infrastructure for AI, designed to collect, organize, govern, and prepare data across domains. Start with language data and build for the information your AI work depends on.
Many domainslanguage data, organization knowledge, research, and learning
Data to AIreviewed datasets and private assistants
A growing collection
Start with language. Build for every domain.
The public catalog begins with TuluEng100, a 100-pair English–Tulu reference dataset. Gena is designed to support data across domains, helping people discover, create, and responsibly reuse collections for different kinds of AI work.
TULUENGLISHTEXT
TuluEng100
100 English–Tulu sentence pairs from the current Hugging Face snapshot. Reviewed by a Tulu-speaking project owner per their report; no independent second review.
Gena is being built as data infrastructure for AI across domains. Language communities, researchers, public institutions, and product teams can structure data for different needs without being limited to one subject or format.
01 / TEXT
Documents and datasets
Bring text, Markdown, CSV, JSON, and JSONL into a workspace, then shape the useful information into a versioned dataset.
02 / AUDIO
Audio and transcripts
Convert spoken content into reviewed transcripts, searchable text, and structured data for later use.
03 / STRUCTURE
Provenance and permissions
Keep sources connected to derived data and make processing, sharing, and model-training choices separately.
Pilot limits: text uploads up to 900 KB; voice clips up to 4 MB per file, with browser recording governed by file size rather than a time cutoff.
Possible use cases
One data platform. Many paths to AI.
Gena’s collection, review, and governance tools can support many kinds of work. These examples show just a few ways people might put the platform to use.
01 / ORGANIZE
Company knowledge
Bring internal decisions, procedures, documents, and meeting notes into a governed knowledge base or reusable dataset.
02 / LEARNING
Course materials
Organize lecture recordings, notes, reading lists, and assignments into learning-focused collections.
03 / RESEARCH
Research and community data
Build domain datasets for language, scientific, and local knowledge projects, with clear sources and reuse terms.
Why Gena exists
The data gap is a language gap.
AI systems are shaped by the data available to build and evaluate them. Many regional and low-resource languages have far less digital data than widely represented languages.
Gena starts by bringing dataset discovery and contribution together, with source, license, and review status visible. The broader goal is data people can understand, evaluate, govern, and responsibly use to build AI.
A platform for the full data journey
Discover. Build. Share with care.
Useful datasets need more than rows. Gena helps people find collections, inspect examples, build new datasets, and understand the conditions for reuse.
01 / DISCOVER
Find and inspect
Search the public catalog, open a dataset, read sample rows, review its license, and download available data.
02 / BUILD
Create in-platform
Start a dataset from structured rows or upload a supported file, then add the details people need to understand it.
03 / SHARE
Publish responsibly
Submissions stay private during review. A configured TuluAI moderator approves a dataset before it enters the public catalog.
From a file to a public dataset
A clear path to contribute.
Contributors keep control of their work. A submission is not public until its rights and license are confirmed and a configured TuluAI moderator approves it.
01
Choose how to start
Upload a CSV, JSON, JSONL, or text file, create structured rows, or contribute a short voice clip.
02
Add context and rights
Describe the language, source, and license, then confirm you have the rights to publish.
03
Save or submit
Keep a private draft, or send the dataset for moderator review before publication.
04
Share after approval
Approved datasets appear in the public catalog with their license and attribution details.
Current limits: text datasets up to 900 KB; voice clips up to 4 MB per file. Browser recording has no fixed time cutoff and stops near the file-size limit.
Before you contribute
Good data needs clear boundaries.
Open a question to learn how privacy, publishing, and future use work in the current pilot.
When does my dataset become public?
It stays private as a draft or while in review. A designated moderator reviews it before it can appear in the public catalog or download area.
What can I do with a public dataset?
Open its details to inspect sample rows, source attribution, and license, then download it for uses permitted by that license. Dataset access does not grant additional rights beyond the listed terms.
Will my dataset automatically be used to train a model?
No. Uploading or publishing a dataset does not automatically grant permission to use it for model development. Any future model-development use needs separate permission.
What kinds of data can Gena support?
Gena is designed as data infrastructure for AI across domains and formats. The current pilot starts with language datasets and private workspaces; selected content sent to an external assistant requires separate consent.
Does fine-tuning have a cost?
Local fine-tuning does not charge TuluAI or require credits. It uses your computer’s processing power, memory, disk, and internet, and the first run downloads training libraries and model weights. The resulting model remains on your computer and is not yet available to Gena’s hosted assistants.
What should I prepare before submitting?
Provide a clear description, language or language pair, source when available, and an explicit license. You must confirm you have the right to publish the data.
Can I contribute voice recordings?
Yes. Contributors must be at least 18, and every speaker must be an adult who has given informed consent. Record in your browser without a fixed time cutoff, or upload audio up to 4 MB per clip. Recordings stay private during review; public release and model-development use are separate choices.
Who can contribute?
Anyone with a TuluAI account can start a contribution. Before submitting, make sure you have the rights to share the material and provide its source and license when available.
Can I ask for a contribution to be removed?
Contributors can delete drafts and request removal of submitted or approved material. If a public dataset has already been downloaded, Gena cannot remove those external copies.
Help close the data gap
Build the data foundation for AI.
Discover, create, organize, and govern data, then choose how it can support AI.
