ResearchBenchmarksMulti-task Model Merging on MMLU, SciQA, and GSM8K (test)Multi-task Model Merging on MMLU, SciQA, and GSM8K (test)Follow74.3Average (Individual)NeuroLoRA71.38872.14472.973.656Mar 12, 2026Evaluation resultsMethodMethodLinksAverage (Individual)Task Arithmetic ScoreTIES ScoreNeuroLoRA2026.0374.371.872.3FlyLoRA2026.0372.668.969.6DoRA2026.0372.259.861.7LoRA2026.0371.558.260.5