Developing LLM-driven agents for reasoning, planning, and long-horizon decision-making in interactive text-based environments.
LLM Agents
Planning
Benchmarking higher-order mathematical problem-solving skills in Large Language Models using Odyssey Math Data.
Dataset
Reasoning
Building autonomous agents capable of navigating and competing in 3D first-person visually complex environments using Reinforcement Learning.
Embodied AI
Vision-RL
Advancing core algorithmic solutions to tackle sparse rewards and sample inefficiency in complex, multi-objective tasks.
RL Algorithms
Control