Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

PSO-Merging: Merging Models Based on Particle Swarm Optimization

About

Model merging has emerged as an efficient strategy for constructing multitask models by integrating the strengths of multiple available expert models, thereby reducing the need to fine-tune a pre-trained model for all the tasks from scratch. Existing data-independent methods struggle with performance limitations due to the lack of data-driven guidance. Data-driven approaches also face key challenges: gradient-based methods are computationally expensive, limiting their practicality for merging large expert models, whereas existing gradient-free methods often fail to achieve satisfactory results within a limited number of optimization steps. To address these limitations, this paper introduces PSO-Merging, a novel data-driven merging method based on the Particle Swarm Optimization (PSO). In this approach, we initialize the particle swarm with a pre-trained model, expert models, and sparsified expert models. We then perform multiple iterations, with the final global best particle serving as the merged model. Experimental results on different language models show that PSO-Merging generally outperforms baseline merging methods, offering a more efficient and scalable solution for model merging.

Kehao Zhang, Shaolei Zhang, Yang Feng• 2025

Related benchmarks

TaskDatasetResultRank
Language UnderstandingMMLU (test)--
167
Mathematical ReasoningGSM8K (val)
Accuracy49
108
Multitask Language UnderstandingMMLU (val)
Accuracy63.5
94
Multi-task Language UnderstandingMMLU (test)
Normalized Accuracy53.9
87
Mathematical ReasoningGSM8K (test)
Accuracy (ACC)26.1
62
Common Sense ReasoningHELLASWAG (test)
Accuracy58.7
56
Commonsense ReasoningHellaSwag (val)
Accuracy65
54
Natural Language UnderstandingGLUE
SST-291.1
40
Truthfulness EvaluationTruthfulQA (test)--
30
Knowledge ReasoningK-Cross (val)
Accuracy43
22
Showing 10 of 24 rows

Other info

Follow for update