Amuro and Char: Analyzing the Relationship between Pre-Training and Fine-Tuning of Large Language Models

TL;DR

This paper investigates how pre-training and fine-tuning influence large language models, revealing that continual pre-training enhances latent capabilities, fine-tuning can cause knowledge forgetting, and more pre-training reduces prompt sensitivity.

Contribution

It provides a comprehensive analysis of the relationship between pre-training and fine-tuning, highlighting the benefits and drawbacks of continual pre-training and fine-tuning strategies.

Findings

Continual pre-training improves model capabilities in a latent manner.

Extra fine-tuning benefits datasets where the model initially underperforms.

Fine-tuning can lead to forgetting of previously learned domain knowledge.

Abstract

The development of large language models leads to the formation of a pre-train-then-align paradigm, in which the model is typically pre-trained on a large text corpus and undergoes a tuning stage to align the model with human preference or downstream tasks. In this work, we investigate the relationship between pre-training and fine-tuning by fine-tuning multiple intermediate pre-trained model checkpoints. Our results on 18 datasets suggest that i) continual pre-training improves the model in a latent way that unveils after fine-tuning; ii) with extra fine-tuning, the datasets that the model does not demonstrate capability gain much more than those that the model performs well during the pre-training stage; iii) although model benefits significantly through supervised fine-tuning, it may forget previously known domain knowledge and the tasks that are not seen during fine-tuning; iv) the model resembles high sensitivity to evaluation prompts after supervised fine-tuning, but this sensitivity can be alleviated by more pre-training.