princeton-nlp/SimPO

Reference implementation for DPO (Direct Preference Optimization)

AllenAI's post-training codebase

Robust recipes to align language models with human and AI preferences

Train transformer language models with reinforcement learning.

Implementation of the training framework proposed in Self-Rewarding Language Model, from MetaAI

A very simple GRPO implement for reproducing r1-like LLM thinking.

Code and documentation to train Stanford's Alpaca models, and generate the data.

Post-training with Tinker

Sky-T1: Train your own O1 preview model within $450