logo
Perpetual Datasets for AI
logo
problem
GenAI models require a ton of labelled data and web2 platforms can't keep up
A handful of companies own and control all the large scale datasets
centralized
Solutions
Enter Perpetual Datasets
The 100x solution offering large scale datasets with universal accessibility and community ownership
logo
Ellipse
logo

Perpetually Labelled

Datasets are created proactively based on expected future demand and licensed for training AI models across multiple instances. Trustless verfification mechanisms ensure absolute absence of bias, fostering reliability and integrity in AI models

This proactive model of labelling ensures that datasets are :
perpetual-labelled-icons-background
Dataset Icon
Universally Accessible
For trustless contributions and Usage
Dataset Icon
Produced at Scale
Increasingly affordable, pay as you go
Dataset Icon
Community owned
Tokenized ownership for licensing revenue
logo
Made Possible by our Novel Protocol
cube
pyramid
cylinder
Contributors
Submit, label and verify data - grow perpetual datasets and earn rewards
Stakers
Secure submissions on the platform by Staking and earn rewards
Data-Owners
Own the Revenue and Governance Rights
fractionai logo
Powered by Frac Token
Ensuring Data Quality
Reputation Based Consensus
testimonials-background
Testimonials
testimonial
Suvam Patra
Research Scientist, Meta Al
block-quote
"I was blown away by the extensive library of labelled datasets. This is like a candy store for anyone working in Al"
logo
Ecosystem Partners
Slide 0
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Slide 0
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
logo
Blog Left Curve
Blog
Slide 0
The day is D-1, one of the most anticipated AI products is about to launch. The suspense h...
Read it
Arrow
Slide 1
You log in to your favourite AI tool to retouch some family photos and all of the sudden.....
Read it
Arrow
Slide 2
As I write this very piece, I’m aware that the first draft will go to the trash (one that ...
Read it
Arrow
Slide 3
When it comes to Artificial Intelligence, we are standing right at the tipping point. One ...
Read it
Arrow
logo
FAQ
background
What is Fraction Al about?

Fraction AI is a decentralized platform where humans and agents work together to create highest quality labelled datasets for training AI models. Be it image, text, audio, video, we do it all. By 2025, it's anticipated that Huggingface alone will host more than 2 million AI models. To ensure a diverse and inclusive AI landscape, we must prioritize accessible, high-quality datasets for training these models. Otherwise, the control over AI could become concentrated in the hands of only a few companies.

What types of datasets are available on the platform?

We support development of datasets encompassing diverse data formats such as text, images, audio, and video, catering to a wide array of AI applications. Within these domains, we focus on various use cases like annotation, bounding box, segmentation, and more. Anyone can initiate a new dataset and contribute to existing ones in a completely trustless manner

How can I participate?

1. Contribute: Choose a dataset of your choice and make submissions or label data based on requirements
2. Verification: Verify contributions and ensure quality of datasets
3. Stake: Frac tokens need to be staked for contribution and verification, delegate your tokens and yield
4. Data License: Buy license to use dataset for commercial purposes
5. Revenue Rights Certificates (Coming soon): Buy rights to get a portion of dataset's licensing revenue

What if I am intersted in a dataset that's not already there?

You're welcome to launch a new dataset on our platform using our Protocol, which operates entirely on a trustless basis. With thousands of contributors motivated by a free-market economy, your dataset can thrive. Additionally, we're pleased to provide funding for datasets that hold promise for benefiting the wider community

How does Blockchain come into the picture?

Well there are several reasons for that:
1. Transparency: Using blockchain enables us to maintain the entire data generation process in the public domain, ensuring the highest quality of generated data
2. Global Participation: By allowing anyone, regardless of location, to participate in the data generation process, we can gather a diverse range of data points. This eliminates the need for extensive vendor contracts and mitigates the numerous regulatory issues that arise across different countries
3. Freedom from Bias and Censorship: It's imperative that AI remains free from the biases and narratives of any particular group. Blockchain-powered verification plays a crucial role in ensuring this neutrality

What sets Fraction AI apart from other data providers?

Other data providers primarily operate as data labeling companies. They require users to supply their own data, which is then labeled according to specific needs. However, most AI model developers lack extensive unlabeled datasets. Our vision is to democratize access to high-quality labeled datasets, making them both affordable and accessible to all. Moreover, we aim to distribute a fair share of the value generated back to the contributors and verifiers. This approach fosters fairness, impartiality, and accessibility of AI for everyone

twitter iconlinkedin icon
logo