Understanding LLM Embeddings for Regression
This work addresses the application of LLMs to regression problems, offering insights for researchers and practitioners in machine learning, though it appears incremental as it builds on existing embedding methods.
The paper tackles the problem of using large language model (LLM) embeddings for regression tasks, demonstrating that these embeddings outperform traditional feature engineering in high-dimensional settings, with performance partly explained by their inherent Lipschitz continuity over numeric data.
With the rise of large language models (LLMs) for flexibly processing information as strings, a natural application is regression, specifically by preprocessing string representations into LLM embeddings as downstream features for metric prediction. In this paper, we provide one of the first comprehensive investigations into embedding-based regression and demonstrate that LLM embeddings as features can be better for high-dimensional regression tasks than using traditional feature engineering. This regression performance can be explained in part due to LLM embeddings over numeric data inherently preserving Lipschitz continuity over the feature space. Furthermore, we quantify the contribution of different model effects, most notably model size and language understanding, which we find surprisingly do not always improve regression performance.