Unverified50% confidenceFactExact time
Al-Dahle's team was responsible for the complete pipeline from pre-training data filtering and model architecture design to safety alignment (RLHF/RLAIF).
1
Sources
50%
Confidence
Long-term
Relevance
10/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedEleutherAI 的 Pythia 项目同时公开了训练数据、代码和完整检查点序列，因此在学术研究中被广泛用于机制性分析61% similarUnverified高质量标注数据是监督学习和RLHF的核心原料，该环节目前仍高度依赖人工完成61% similarUnverifiedGarak focuses on LLM security probing, Promptfoo is a prompt testing framework, and Microsoft PyRIT is a red-teaming automation toolkit.60% similarUnverified《AI行政令》要求训练算力超过10²⁶次浮点运算（FLOP）的模型开发商向联邦政府报告安全测试结果60% similarUnverifiedMainstream security alignment methods include RLHF, Constitutional AI, and red-teaming.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/967752API
curl https://kongchang.com/api/v1/knowledge/claims/967752MCP
get_claim(id=967752)