Language models
Structured Alignment LM
Plan-first LLM alignment
Fine-tunes TinyLlama for structured plan-first generation with LoRA supervised fine-tuning and PPO-based reinforcement learning.
Project scope
A plan-first language-model alignment project with a FastAPI backend and a Next.js interface for comparing aligned generation with a baseline.
Explore the implementation
Source code, setup instructions, and project documentation are available in the repository.