Saved in:
Bibliographic Details
Main Authors: Sudalairaj, Shivchander, Bhandwaldar, Abhishek, Pareja, Aldo, Xu, Kai, Cox, David D., Srivastava, Akash
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2403.01081
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929331293388800
author Sudalairaj, Shivchander
Bhandwaldar, Abhishek
Pareja, Aldo
Xu, Kai
Cox, David D.
Srivastava, Akash
author_facet Sudalairaj, Shivchander
Bhandwaldar, Abhishek
Pareja, Aldo
Xu, Kai
Cox, David D.
Srivastava, Akash
contents This work introduces LAB (Large-scale Alignment for chatBots), a novel methodology designed to overcome the scalability challenges in the instruction-tuning phase of large language model (LLM) training. Leveraging a taxonomy-guided synthetic data generation process and a multi-phase tuning framework, LAB significantly reduces reliance on expensive human annotations and proprietary models like GPT-4. We demonstrate that LAB-trained models can achieve competitive performance across several benchmarks compared to models trained with traditional human-annotated or GPT-4 generated synthetic data. Thus offering a scalable, cost-effective solution for enhancing LLM capabilities and instruction-following behaviors without the drawbacks of catastrophic forgetting, marking a step forward in the efficient training of LLMs for a wide range of applications.
format Preprint
id arxiv_https___arxiv_org_abs_2403_01081
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LAB: Large-Scale Alignment for ChatBots
Sudalairaj, Shivchander
Bhandwaldar, Abhishek
Pareja, Aldo
Xu, Kai
Cox, David D.
Srivastava, Akash
Computation and Language
Machine Learning
This work introduces LAB (Large-scale Alignment for chatBots), a novel methodology designed to overcome the scalability challenges in the instruction-tuning phase of large language model (LLM) training. Leveraging a taxonomy-guided synthetic data generation process and a multi-phase tuning framework, LAB significantly reduces reliance on expensive human annotations and proprietary models like GPT-4. We demonstrate that LAB-trained models can achieve competitive performance across several benchmarks compared to models trained with traditional human-annotated or GPT-4 generated synthetic data. Thus offering a scalable, cost-effective solution for enhancing LLM capabilities and instruction-following behaviors without the drawbacks of catastrophic forgetting, marking a step forward in the efficient training of LLMs for a wide range of applications.
title LAB: Large-Scale Alignment for ChatBots
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2403.01081