The Gretel Epoch #7
Hi everyone, it’s Ali from Gretel.ai, sharing our latest newsletter here which highlights a few initiatives we’re really excited about:
Product
At Gretel, we’re taking a ‘multi-model’ approach to synthetic data generation. Whatever your use case or data type, our goal is to be a one-stop-shop for high-quality synthetic data. That’s why we just released support for two new state-of-the-art generative synthetic models– DoppelGANger offers unparalleled generation quality for complex time-series data such as medical sensors and stock data. Gretel GPT is an open-source implementation of the GPT-3 architecture that makes it simple to augment datasets to train chat-bots and sentiment classifiers or to generate your own text examples from scratch. Check out our Jupyter notebook examples to get started!
Education
Our team is building a platform to help users create, share and build ecosystems around data. We’ve done a ton of work already to make Gretel a no-code tool that anyone can use. But the other big piece is education — explaining what synthetic data is, why it’s necessary and how it can solve a host of critical data challenges.
To demystify the concept a bit, we launched our first learning pathway. It’s a curation of our best resources on the core concepts of synthetic data generation. You can learn at your own pace, and by the end of it, you’ll be generating artificial data that’s even better than the real thing. :-)
And if you’re interested to do a deeper dive, you can also check out our new GitHub repo of other helpful synthetic data resources. Send a pull request if you want to contribute!
Community
Lastly, we are focused on learning in public, alongside our users, in some cases. For example, when we hear about new modeling techniques or someone flags an interesting idea in our Slack community, our Applied Research team is quick to test and share their own results. That’s the genesis of blogs like diffusion models for document synthesis, model soup, and ’red teaming’ synthetic models.
We love engaging with users. Whether it’s through hosting a live AMA on our new time-series model, talking to emergency medicine physicians about powering production-grade AI predictions in healthcare or even traveling to Helsinki to discuss the effects of sampling procedures on the quality of synthetic tabular data. We strive to meet developers where they are.
Each of these initiatives was sparked by conversations with users, like you, in our community. So on behalf of our team, thank you for all the feedback and continued support. If you ever have any questions, feel free to reach out to us at hi@gretel.ai or join our Slack community and share your thoughts with our staff. Also, if you’re passionate about joining a diverse team that’s helping drive ethical AI innovations, we’re always hiring.
Thanks again and happy synthesizing.
Ali Golshan
Co-founder/CEO
