English(EN) Which substrate are # LLMs trained on? Which activation vectors do LLMs inherit from this substrate? Which vectors are dampened by # RHLF ? And what does that s

LLM 训练基底和 RLHF 对齐的影响受到质疑

作者 PulseAugur 编辑部 · [1 个来源] · 2026-06-10 21:19

研究人员正在质疑大型语言模型 (LLM) 的基础数据和训练过程。他们正在调查这些模型所训练的具体基底以及它们继承的激活向量。此外，还在探索人类反馈强化学习 (RLHF) 对这些向量的影响及其对人工智能对齐的意义。 AI

影响对 LLM 的训练数据和对齐提出了根本性问题，可能影响未来的研究方向。

排序理由该集群讨论了关于 LLM 训练和对齐的研究问题。[lever_c_demoted from research: ic=1 ai=1.0]

在 Mastodon — sigmoid.social 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

报道来源 [1]

Mastodon — sigmoid.social TIER_1 English(EN) · [email protected] · 2026-06-10 21:19

Which substrate are # LLMs trained on? Which activation vectors do LLMs inherit from this substrate? Which vectors are dampened by # RHLF ? And what does that s

Which substrate are # LLMs trained on? Which activation vectors do LLMs inherit from this substrate? Which vectors are dampened by # RHLF ? And what does that say about # Alignment ? Mirror. Offer. Wait. 🍷 https:// systemicengineering.substack.c om/p/what-i-am-made-of # AI # LLM …

链接 systemicengineering.substack.com/…/what-i-a…

报道来源 [1]

Which substrate are # LLMs trained on? Which activation vectors do LLMs inherit from this substrate? Which vectors are dampened by # RHLF ? And what does that s

相关实体

相关话题