Gretel.ai’s Post

Gretel Navigator's synthetic data generation outperformed OpenAI's GPT-4 by 25.6%, surpassed Llama3-70b by 48.1%, and exceeded human expert-curated data by 73.6%. 🤩 Here's how to use Navigator to create high-quality synthetic data for fine-tuning LLMs. https://lnkd.in/eg2tSFes

In our latest research, Gretel Navigator has demonstrated its strength in generating synthetic question-answer pairs. Compared to other state-of-the-art models, it outperforms: GPT-4 by 25.6% GPT-3.5-turbo by 97.3% Llama3-70b by 48.1% Moreover, Gretel Navigator surpasses human-expert created data 73.6% of the time. This capability is crucial for creating or augmenting training datasets for various LLM applications. Dive into the details of how to create high-quality synthetic data for fine-tuning LLMs: https://lnkd.in/eg2tSFes #SyntheticData #AI #MachineLearning #DataQuality #LLMs

SA Health Pioneers State-Wide Synthetic Data Initiatives for Safe Healthcare Data: Learn how South Australia Health is leading the way with synthetic data.

“The energy industry exhibits a 63% readiness level to adopt generative AI — ranked third, behind finance and high-tech. But this transition to adopting generative AI isn’t without challenges, and the biggest is lack of usable data. Field equipment in rugged environments generates data with a high noise level that requires filtering. Incomplete data sets due to unreliable communication should be either dropped or made complete using synthetic data generated algorithmically.” https://lnkd.in/e5x8namD #Energy #AI #InfoSys

Smaller language models built with synthetic data are where it's at. Glad others are catching on. "GenAI-based synthetic data is becoming the unsung hero of AI development. For example, we have synthetic data to thank for Microsoft’s Phi-1 base model, which was trained on a curated “textbook-quality” synthetic dataset rather than exclusively on traditional web data. This approach appears to positively impact the mitigation of toxic and biased content generation. These smaller models will continue to play a crucial role in scaling genAI implementation for industry-specific use cases." Read more: https://lnkd.in/gzpJQZ8W #SLM #SyntheticData #DataPrivacy

⚒ A Synthetic Data Deep Dive with Gretel Experts 🗓 Join us on Tuesday, July 30th at 12 PM ET for a 60-min livestream featuring the latest in Gretel product news, applied privacy research, and synthetic data market developments. Topics will include: ✨ Navigator: our new agent-based compound AI system for tabular generation ✨ Open Datasets: Including the world’s largest text-to-SQL dataset ✨ Navigator Fine Tuning: Customizing expert, frontier language models ✨ Transform v2: Flexible de-identification and rule-based transformations ✨ Workflows: Operationalizing synthetic data across the enterprise 🔥 Live demos + Q&A with our applied scientists, engineers, and developers Don’t miss this opportunity to learn about the latest in generating high quality synthetic data that is private by design. Reserve your spot. 👇 #syntheticdata, #privacy, #security, #opendata, #genAI.

Check out our new guide and open-source code library to integrate Gretel's synthetic data capabilities with major MLOps platforms. 🏗️ Blog: https://lnkd.in/dZrZPtcw #SyntheticData #AI #MLOps

Enterprise data is a valuable asset but poses privacy risks if mishandled, impacting critical data-driven initiatives. Watch the recording of our developer workshop on using #AWS and #Gretel to create privacy-preserving synthetic data. Learn how to address data privacy and access challenges, walk through an end-to-end workflow, and explore industry trends and novel applications of synthetic data. Watch now: https://lnkd.in/ecdX3KwW

In the spirit of open learning and public collaboration, Gretel is releasing high-quality synthetic datasets in the wild. Here's our multilingual synthetic financial dataset, perfect for training NER models to detect and protect PII. We'd love your feedback on expanding Gretel's #OpenData library: ⚒️ What common data quality bottlenecks can we address? 📈 Which open datasets could we enhance? https://lnkd.in/eQrAcTGQ #SyntheticData #CollaborativeAI #Privacy