Instruction fine-tuning on AI Research Instructions and Outputs
87.5Accuracy (%)Proposed Method
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Proposed MethodTrainable Params (%)=0.62025.12 | 87.5 | 7.4 | |
| DP-LoRATrainable Params (%)=12025.12 | 86.2 | 7.6 | |
| AnaDPTrainable Params (%)=0.52025.12 | 85.8 | 7.8 | |
| DP-ForwardTrainable Params (%)=0.32025.12 | 85 | 7.9 | |
| DP-BiTFiTTrainable Params (%)=0.12025.12 | 84.5 | 8 |