Natural Language Understanding on GLUE (imbalanced partition)
94.7248SST-2 AccuracyCENTRALISED
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CENTRALISEDAggregation Method=CENTRALISED, Pre-trained Model=RoBERTa-base, Parameter-Efficient Fine-Tuning=LoRA, Data Distribution=imbalanced2025.01 | 94.7248 | 86.0316 | |
| FedAvgAggregation Method=FEDAVG, Pre-trained Model=RoBERTa-base, Parameter-Efficient Fine-Tuning=LoRA, Data Distribution=imbalanced2025.01 | 93.1193 | 85.7769 | |
| FRA-LoRAAggregation Method=FRA, Pre-trained Model=RoBERTa-base, Parameter-Efficient Fine-Tuning=LoRA, Data Distribution=imbalanced2025.01 | 92.7752 | 85.5018 | |
| FFAAggregation Method=FFA, Pre-trained Model=RoBERTa-base, Parameter-Efficient Fine-Tuning=LoRA, Data Distribution=imbalanced2025.01 | 92.5459 | 83.5354 |