Personalized Federated Learning on DRAKE alpha=1.0 Dirichlet non-i.i.d. splits
69.12A_lastFedMosaic
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FedMosaicModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 69.12 | 64.72 | |
| FedITModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 66.89 | 62.57 | |
| SFTModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 66.77 | 62.38 | |
| TAKFLModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 65.18 | 63.57 | |
| FedMKTModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 64.87 | 60.67 | |
| FedDPAModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 64.34 | 60.34 | |
| FedSIMModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 63.73 | 59.82 | |
| FedDATModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 62.99 | 59.19 | |
| DITTOModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 62 | 58.37 | |
| PerADAModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 61.73 | 58.2 |