Personalized Federated Learning on DRAKE alpha=0.1 Dirichlet non-i.i.d. splits
73.12A_lastFedMosaic
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FedMosaicModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 73.12 | 68.39 | |
| FedITModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 72.07 | 67.17 | |
| SFTModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 71.4 | 66.9 | |
| FedSIMModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 69.9 | 65.03 | |
| TAKFLModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 69.88 | 65.21 | |
| FedMKTModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 69.04 | 64.78 | |
| FedDPAModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 68.56 | 64.58 | |
| DITTOModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 66.8 | 63.46 | |
| FedDATModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 66.43 | 63.16 | |
| PerADAModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 66.41 | 63.27 |