Gifted
PHD, COMPUTER SCIENCE, UNIV OF SOUTHERN CALIFORNIA
Kiran Lekkala
Super-power:
Reinforcement learning, self-supervised learning, pretraining, LLMs
Interests:
Large-scale self-supervised/unsupervised pretraining, task-specific fine-tuning, and transfer learning
Why AGD?
I am passionate about AGD because it’s the need of the hour. LLMs can generate text, but without mimicking complex human behaviors and decisions, we can’t scale these models to achieve human-level general intelligence.