Verified70% confidenceFactTime unknown

OpenAI和Anthropic都采用了RLHF和代码执行反馈来优化模型的编程表现

4

Sources

70%

Confidence

Long-term

Relevance

6/3/2026

First Seen

Sources

OpenAI Codex越来越像Claude Code？开发者吐槽产品同质化

bilibiliwunanyx

Related Entities

OpenAIcompany Anthropiccompany RLHFconcept

Related Claims

UnverifiedOpenAI通过Agent Kit的Evaluate模块收集用户好评和差评数据，可直接用于RLHF训练以提升模型表现80% similar UnverifiedRLHF 最初由 OpenAI 用于语言模型的对齐训练，后被广泛应用于图像生成领域78% similar UnverifiedRLHF 由 OpenAI 于 2017 年前后系统化，最初用于让模型输出更符合人类价值观76% similar UnverifiedOpenAI采用RLHF（基于人类反馈的强化学习）和规则层过滤器来拦截敏感请求76% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/39410

API

curl https://kongchang.com/api/v1/knowledge/claims/39410

MCP

get_claim(id=39410)