Research
Our team is conducting industry leading research
Welcome to our research page, where we delve into our work on foundational principles of generative deep learning architectures and algorithms, focusing primarily on Large Language Models (LLMs). Our research is dedicated to unraveling the intricate mechanics of these systems, drawing inspiration from deep network geometry and insights from neuroscience.
HOW WE DO IT
Here's a glimpse into our ongoing research initiatives
Transfer Learning in Generative AI
We are developing innovative algorithms that facilitate seamless transfer learning of LLMs across various verticals. By addressing challenges such as the mitigation of catastrophic forgetting phenomena, we aim to unlock the full potential of transfer learning in LLMs.
Trustworthy Generative AI
Ensuring the safe and reliable deployment of enterprise-grade LLMs is paramount to our customers. We are actively developing diverse guardrails and solutions that enhance the trustworthiness of these models, enabling their effective integration into real-world applications.
Resource-Efficient Generative AI
We are focusing on optimizing ML architectures to improve efficiency and reduce inference computing and memory costs. By streamlining the computational process and understanding the nature of the learned representations, we aim to enhance the scalability and practicality of generative AI solutions. At Tenyx, we are committed to pushing the boundaries of generative AI research, driving innovation, and paving the way for the next generation of intelligent systems.
January 11, 2024
Scaling Laws for Forgetting When Fine-Tuning Large Language Models
December 4, 2023
Characterizing Large Language Model Geometry Solves Toxicity Detection and Generation
November 28, 2023