Srijith-rkr/iterative-alignment

stars

commits

Python

primary language

Oct 13, 2024

updated

README

training Contains training pipelines for Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO). Note: The Self-Play training pipeline is still under development.
inference_and_scoring contains the modified implementaion from https://github.com/hkust-nlp/deita to score samples on quality and complexity. Added multi-GPU inference support using Hugging Face’s accelerate library.
To reproduce the environment, use the provided environment.yml file

5 commits

cy0307/lm-dpo-alignment

zyttt-coder/SIPO

Official code for paper "Self-Improvement Towards Pareto Optimality: Mitigating Preference…

ZurichNLP/HF4ATS

Code and data for HF4ATS

yuanyige/sea

[NeurIPS 2025] Inference-time Alignment in Continuous Space

uclaml/SPPO

The official implementation of Self-Play Preference Optimization (SPPO)

589

architsharma97/dpo-rlaif

100

princeton-nlp/unintentional-unalignment

[ICLR 2025] Unintentional Unalignment: Likelihood Displacement in Direct Preference Optimization

PKU-Alignment/align-anything

Align Anything: Training All-modality Model with Feedback

4,671

Python

92.8%

Shell

7.2%

README

training Contains training pipelines for Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO). Note: The Self-Play training pipeline is still under development.
inference_and_scoring contains the modified implementaion from https://github.com/hkust-nlp/deita to score samples on quality and complexity. Added multi-GPU inference support using Hugging Face’s accelerate library.
To reproduce the environment, use the provided environment.yml file