desBugger/constitutional-mt

Constitutional midtraining for LLM alignment: values-based document generation, 2×2 factorial training pipeline, and full evaluation suite (including benchmarks on value conflict resolution and sycophantic pressure)

语言：Jupyter Notebook · GitHub 星标：8 · Fork：0

话题：ai-alignment · ai-safety · constitutional-ai · large-language-models · llm-alignment · llm-safety · midtraining

在 GitHub 上查看原仓库