NeMo-Aligner: Scalable Toolkit for Efficient Model Alignment
This provides a practical solution for researchers and developers working on aligning large-scale LLMs, though it is incremental as it builds on existing alignment techniques with optimizations for scalability.
The authors tackled the challenge of aligning large language models with human values by creating NeMo-Aligner, a scalable toolkit that efficiently trains models like Nemotron 4 340B and Llama 3.1 405B on up to a thousand GPUs, supporting multiple alignment paradigms such as RLHF and DPO.
Aligning Large Language Models (LLMs) with human values and preferences is essential for making them helpful and safe. However, building efficient tools to perform alignment can be challenging, especially for the largest and most competent LLMs which often contain tens or hundreds of billions of parameters. We create NeMo-Aligner, a toolkit for model alignment that can efficiently scale to a thousand GPUs for training the largest open-source LLMs such as Nemotron 4 340B and Llama 3.1 405B. NeMo-Aligner comes with highly optimized and scalable implementations for major paradigms of model alignment such as: Reinforcement Learning from Human Feedback (RLHF), Direct Preference Optimization (DPO), SteerLM, and Self-Play Fine-Tuning (SPIN). Additionally, our toolkit supports running most of the alignment techniques in a Parameter Efficient Fine-Tuning (PEFT) setting. NeMo-Aligner is designed for extensibility, allowing support for other alignment techniques with minimal effort. It is open-sourced with Apache 2.0 License and we invite community contributions at https://github.com/NVIDIA/NeMo-Aligner