Eureka: Human-Level Reward Design via Coding Large Language Models

• URL: http://arxiv.org/abs/2310.12931v2
• Date: Tue, 30 Apr 2024 21:35:53 GMT
• Title: Eureka: Human-Level Reward Design via Coding Large Language Models
• Authors: Yecheng Jason Ma, William Liang, Guanzhi Wang, De-An Huang, Osbert Bastani, Dinesh Jayaraman, Yuke Zhu, Linxi Fan, Anima Anandkumar,
• Abstract summary: Large Language Models (LLMs) have excelled as high-level semantic planners for sequential decision-making tasks. We present Eureka, a human-level reward design algorithm powered by LLMs. Eureka exploits the remarkable zero-shot generation, code-writing, and in-context improvement capabilities of state-of-the-art LLMs.
• Score: 121.91007140014982
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large Language Models (LLMs) have excelled as high-level semantic planners for sequential decision-making tasks. However, harnessing them to learn complex low-level manipulation tasks, such as dexterous pen spinning, remains an open problem. We bridge this fundamental gap and present Eureka, a human-level reward design algorithm powered by LLMs. Eureka exploits the remarkable zero-shot generation, code-writing, and in-context improvement capabilities of state-of-the-art LLMs, such as GPT-4, to perform evolutionary optimization over reward code. The resulting rewards can then be used to acquire complex skills via reinforcement learning. Without any task-specific prompting or pre-defined reward templates, Eureka generates reward functions that outperform expert human-engineered rewards. In a diverse suite of 29 open-source RL environments that include 10 distinct robot morphologies, Eureka outperforms human experts on 83% of the tasks, leading to an average normalized improvement of 52%. The generality of Eureka also enables a new gradient-free in-context learning approach to reinforcement learning from human feedback (RLHF), readily incorporating human inputs to improve the quality and the safety of the generated rewards without model updating. Finally, using Eureka rewards in a curriculum learning setting, we demonstrate for the first time, a simulated Shadow Hand capable of performing pen spinning tricks, adeptly manipulating a pen in circles at rapid speed.

Related papers

• Few-shot In-Context Preference Learning Using Large Language Models [15.84585737510038]
  Designing reward functions is a core component of reinforcement learning. It can be exceedingly inefficient to learn rewards as they are often learned tabula rasa. We propose In-Context Preference Learning (ICPL) to accelerate learning reward functions from preferences.
  arXiv  Detail & Related papers  (2024-10-22T17:53:34Z)
• Affordance-Guided Reinforcement Learning via Visual Prompting [51.361977466993345]
  Keypoint-based Affordance Guidance for Improvements (KAGI) is a method leveraging rewards shaped by vision-language models (VLMs) for autonomous RL. On real-world manipulation tasks specified by natural language descriptions, KAGI improves the sample efficiency of autonomous RL and enables successful task completion in 20K online fine-tuning steps.
  arXiv  Detail & Related papers  (2024-07-14T21:41:29Z)
• REvolve: Reward Evolution with Large Language Models using Human Feedback [6.4550546442058225]
  Large language models (LLMs) have been used for reward generation from natural language task descriptions. LLMs, guided by human feedback, can be used to formulate reward functions that reflect human implicit knowledge. We introduce REvolve, a truly evolutionary framework that uses LLMs for reward design in reinforcement learning.
  arXiv  Detail & Related papers  (2024-06-03T13:23:27Z)
• Learning Reward for Robot Skills Using Large Language Models via Self-Alignment [11.639973274337274]
  Large Language Models (LLM) contain valuable task-related knowledge that can potentially aid in the learning of reward functions. We propose a method to learn rewards more efficiently in the absence of humans.
  arXiv  Detail & Related papers  (2024-05-12T04:57:43Z)
• Enhancing Q-Learning with Large Language Model Heuristics [0.0]
  Large language models (LLMs) can achieve zero-shot learning for simpler tasks, but they suffer from low inference speeds and occasional hallucinations. We propose textbfLLM-guided Q-learning, a framework that leverages LLMs as hallucinations to aid in learning the Q-function for reinforcement learning.
  arXiv  Detail & Related papers  (2024-05-06T10:42:28Z)
• Instructed Language Models with Retrievers Are Powerful Entity Linkers [87.16283281290053]
  Instructed Generative Entity Linker (INSGENEL) is the first approach that enables casual language models to perform entity linking over knowledge bases. INSGENEL outperforms previous generative alternatives with +6.8 F1 points gain on average.
  arXiv  Detail & Related papers  (2023-11-06T16:38:51Z)
• Learning to Optimize for Reinforcement Learning [58.01132862590378]
  Reinforcement learning (RL) is essentially different from supervised learning, and in practice, these learneds do not work well even in simple RL tasks. Agent-gradient distribution is non-independent and identically distributed, leading to inefficient meta-training. We show that, although only trained in toy tasks, our learned can generalize unseen complex tasks in Brax.
  arXiv  Detail & Related papers  (2023-02-03T00:11:02Z)
• Example-Driven Model-Based Reinforcement Learning for Solving Long-Horizon Visuomotor Tasks [85.56153200251713]
  We introduce EMBR, a model-based RL method for learning primitive skills that are suitable for completing long-horizon visuomotor tasks. On a Franka Emika robot arm, we find that EMBR enables the robot to complete three long-horizon visuomotor tasks at 85% success rate.
  arXiv  Detail & Related papers  (2021-09-21T16:48:07Z)
• Emergent Real-World Robotic Skills via Unsupervised Off-Policy Reinforcement Learning [81.12201426668894]
  We develop efficient reinforcement learning methods that acquire diverse skills without any reward function, and then repurpose these skills for downstream tasks. We show that our proposed algorithm provides substantial improvement in learning efficiency, making reward-free real-world training feasible. We also demonstrate that the learned skills can be composed using model predictive control for goal-oriented navigation, without any additional training.
  arXiv  Detail & Related papers  (2020-04-27T17:38:53Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.