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

Eureka: Human-Level Reward Design via Coding Large Language Models — mentioned by stalfie

Thu, 07 May 2026 15:30:25 GMT

Well, if the evaluation infrastructure is something humans could have had access to before, and that the agents key "skill" is just that it's a more patient and scalable worker, I would still argue that this "comes from the agent". Humans get bored, inpatient, or run out of time, and so often give up in what they perceive to be a decent "local minima". Early verification harnesses using gpt-4 for optimizing robot reward functions succeeded quite well on the fact that the LLM just kept going (link below). As long as it is too boring for a human to use the same evaluation infrastructure, this…

Paper and discussion history. A mention is not an endorsement.

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

Wed, 25 Oct 2023 12:09:56 GMT

Eureka: Human level reward design via coding large language models

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

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

Mon, 23 Oct 2023 14:06:23 GMT

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

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Eureka: Human-Level Reward Design via Coding Large Language Models — mentioned by famouswaffles

Sat, 21 Oct 2023 16:30:43 GMT

From the paper, https://arxiv.org/abs/2310.12931 "In addition, by examining the average correlation by task (App. E), we observe that the harder the task is, the less correlated the EUREKA rewards. We hypothesize that human rewards are less likely to be near optimal for difficult tasks, leaving more room for EUREKA rewards to be different and better. In a few cases, EUREKA rewards are even negatively correlated with human rewards but perform significantly better, demonstrating that EUREKA can discover novel reward design principles that may run counter to human intuition"

Paper and discussion history. A mention is not an endorsement.

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

Sat, 21 Oct 2023 16:09:03 GMT

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

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.