Unverified50% confidenceFactExact time
Constitutional AI training is divided into two stages: in the first stage, the model critiques and revises its own outputs according to constitutional principles, and in the second stage, RLHF training is conducted using the revised data.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAnthropic's flagship Constitutional AI safety framework divides training into two phases: SLAIF self-evaluation revision and RLAIF reinforcement learning.76% similarUnverifiedCAI训练分为两阶段：先让模型依据原则进行自我批评与修正（SL-CAI），再通过基于AI反馈的强化学习（RLAIF）优化72% similarUnverified美国AI监管以模型训练算力阈值（以FLOP为单位）作为触发强制报备义务的量化指标68% similarUnverifiedLLaVA的训练分为两个阶段：第一阶段是预训练对齐（冻结视觉编码器和语言模型，仅训练投影层），第二阶段是指令微调（解冻语言模型进行端到端训练）67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/895424API
curl https://kongchang.com/api/v1/knowledge/claims/895424MCP
get_claim(id=895424)