Knowledge Fusion By Evolving Weights of Language Models
This addresses the need for efficient and generalizable language models across diverse domains, though it is incremental as it builds on existing model merging frameworks.
The paper tackles the problem of fine-tuning language models requiring extensive resources and varying performance across domains by proposing Evolver, a knowledge fusion method that integrates multiple models into a unified one without further training, achieving large-margin improvements over previous state-of-the-art models on mainstream language models.
Fine-tuning pre-trained language models, particularly large language models, demands extensive computing resources and can result in varying performance outcomes across different domains and datasets. This paper examines the approach of integrating multiple models from diverse training scenarios into a unified model. This unified model excels across various data domains and exhibits the ability to generalize well on out-of-domain data. We propose a knowledge fusion method named Evolver, inspired by evolutionary algorithms, which does not need further training or additional training data. Specifically, our method involves aggregating the weights of different language models into a population and subsequently generating offspring models through mutation and crossover operations. These offspring models are then evaluated against their parents, allowing for the preservation of those models that show enhanced performance on development datasets. Importantly, our model evolving strategy can be seamlessly integrated with existing model merging frameworks, offering a versatile tool for model enhancement. Experimental results on mainstream language models (i.e., encoder-only, decoder-only, encoder-decoder) reveal that Evolver outperforms previous state-of-the-art models by large margins. The code is publicly available at {https://github.com/duguodong7/model-evolution}.