Repurposing Language Models into Embedding Models: Finding the Compute-Optimal Recipe
This provides a practical guide for practitioners designing embedding models, but it is incremental as it builds on existing fine-tuning methods.
The paper tackles the problem of compute-optimal training for text embedding models from pre-trained language models, resulting in a recipe that identifies full fine-tuning as optimal at lower budgets and low-rank adaptation at higher budgets.
Text embeddings are essential for many tasks, such as document retrieval, clustering, and semantic similarity assessment. In this paper, we study how to contrastively train text embedding models in a compute-optimal fashion, given a suite of pre-trained decoder-only language models. Our innovation is an algorithm that produces optimal configurations of model sizes, data quantities, and fine-tuning methods for text-embedding models at different computational budget levels. The resulting recipe, which we obtain through extensive experiments, can be used by practitioners to make informed design choices for their embedding models. Specifically, our findings suggest that full fine-tuning and low-rank adaptation fine-tuning produce optimal models at lower and higher computational budgets respectively.