Shypd.aiThe largest open index of AI toolsEureka is an open-source research tool that enables human-level reward design for reinforcement learning using large language models. It automates the generation and optimization of reward code, outperforming human experts in many tasks.
Trending
About
Eureka is an open-source research tool designed to bridge the gap in using large language models (LLMs) for complex low-level manipulation tasks in reinforcement learning. It leverages the zero-shot generation, code-writing, and in-context improvement capabilities of state-of-the-art LLMs like GPT-4 to perform evolutionary optimization over reward code. This allows Eureka to generate reward functions that surpass expert human-engineered rewards without task-specific prompting or predefined templates. The tool has demonstrated superior performance across 29 open-source RL environments, outperforming human experts on 83% of tasks with an average normalized improvement of 52%. Eureka also supports a gradient-free approach to reinforcement learning from human feedback (RLHF) and has enabled a simulated five-finger Shadow Hand to perform pen spinning tricks.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
