Personalized Federated Learning on DRAKE Dirichlet non-i.i.d. splits alpha=0.5
70.2A_lastFedMosaic
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FedMosaicModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 70.2 | 64.68 | |
| SFTModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 68.25 | 63.73 | |
| FedITModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 67.65 | 63.46 | |
| FedSIMModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 67.44 | 62.77 | |
| TAKFLModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 66.65 | 62.08 | |
| FedDPAModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 65.58 | 61.91 | |
| FedMKTModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 65.14 | 60.81 | |
| FedDATModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 63.01 | 59.95 | |
| PerADAModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 62.88 | 59.77 | |
| DITTOModel Setup=4x LLaVA-Llama3.2-1B, 6x LLaVA-Llama3.2-3B2025.05 | 62.34 | 59.52 |