Question Answering on ARC-C (Client Models: GPT-2-Small, OPT-1.3B, LLaMa2-1.3B)
21.8Accuracy (GPT-2-Small)FedCoLLM
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FedCoLLM2024.11 | 21.8 | 29.4 | 30.6 | |
| Standalone2024.11 | 21.4 | 28.2 | 29.3 | |
| FedAvg2024.11 | 20.9 | 28.8 | 29.4 | |
| Zero-Shot2024.11 | 19 | 23.4 | 26.7 |