Physics of Language Models: Part 3.1, Knowledge Storage and Extraction

Published: 02 May 2024, Last Modified: 25 Jun 2024ICML 2024 SpotlightEveryoneRevisionsBibTeXCC BY-NC-ND 4.0
Abstract: Large language models (LLMs) can store a vast amount of world knowledge, often extractable via question-answering (e.g., "What is Abraham Lincoln's birthday?''). However, do they answer such questions based on exposure to similar questions during training (i.e., cheating), or by genuinely learning to extract knowledge from sources like Wikipedia? In this paper, we investigate this issue using a controlled biography dataset. We find a strong correlation between the model's ability to extract knowledge and various _diversity measures_ of the training data. **Essentially**, for knowledge to be reliably extracted, it must be sufficiently augmented (e.g., through paraphrasing, sentence shuffling) _during pretraining_. Without such augmentation, knowledge may be memorized but not extractable, leading to 0% accuracy, regardless of subsequent instruction fine-tuning. To understand why this occurs, we employ (nearly) linear probing to demonstrate a strong connection between the observed correlation and _how the model internally encodes knowledge_ --- whether it is linearly encoded in the hidden embeddings of entity names or distributed across other token embeddings in the training text. **This paper provides several key recommendations for LLM pretraining in the industry:** (1) rewrite the pretraining data --- using small, auxiliary models --- to provide knowledge augmentation, and (2) incorporate more instruction-finetuning data into the pretraining stage before it becomes too late.
Submission Number: 7255
Loading