Tag: model merging
Browse our exclusive articles!
Evo-L2S: Efficient Multi-Objective Model Merging
Discover Evo-L2S, a multi-objective evolutionary merging framework that boosts reasoning model efficiency and accuracy while reducing output length.
Effective Multilingual Model Merging for Machine Translation
Explore challenges and insights in weight-space model merging for multilingual machine translation to improve translation accuracy across diverse languages...
Null-Space Compression for Label-Free LoRA Model Merging
Discover Null-Space Compression, a label-free method for merging LoRA fine-tuned models across classification, regression, and vision-language tasks.
TARA-Merging: Optimized LoRA Merging for AI Models
Discover TARA-Merging, a method to improve LoRA merging by preserving subspace coverage and reducing directional anisotropy for better AI task performance.