Open-domain Question Answering on WebQuestions (WebQS)
8.17WebQS ScoreCorDA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CorDA#Params=320M2024.06 | 8.17 | 19.5 | — | — | |
| FoLoRA#Param=320M, Base Model=LLaMA-2-7B, Fine-tuning Dataset=WizardLM-Evol-Instruct, Evaluation Seeds=32026.05 | 7.97 | 23.71 | 94.44 | 29.26 | |
| LoRA#Params=320M2024.06 | 7.73 | 17.52 | — | — | |
| CorDA#Param=320M, Base Model=LLaMA-2-7B, Fine-tuning Dataset=WizardLM-Evol-Instruct, Evaluation Seeds=32026.05 | 7.28 | 22.37 | 89.13 | 27.76 | |
| OPLoRA#Param=320M, Base Model=LLaMA-2-7B, Fine-tuning Dataset=WizardLM-Evol-Instruct, Evaluation Seeds=32026.05 | 7.08 | 21.39 | 85.21 | 24.75 | |
| LoRA-Null#Param=320M, Base Model=LLaMA-2-7B, Fine-tuning Dataset=WizardLM-Evol-Instruct, Evaluation Seeds=32026.05 | 6.88 | 23.06 | 91.87 | 27.42 | |
| Full fine-tuning#Params=6738M2024.06 | 6.84 | 11.69 | — | — | |
| MiLoRA#Param=320M, Base Model=LLaMA-2-7B, Fine-tuning Dataset=WizardLM-Evol-Instruct, Evaluation Seeds=32026.05 | 6.74 | 21.55 | 85.86 | 26.34 | |
| LoRA#Param=320M, Base Model=LLaMA-2-7B, Fine-tuning Dataset=WizardLM-Evol-Instruct, Evaluation Seeds=32026.05 | 6.34 | 18.94 | 75.45 | 24.8 | |
| LLaMA-2-7BBase Model=LLaMA-2-7B, Fine-tuning Dataset=WizardLM-Evol-Instruct, Evaluation Seeds=32026.05 | 5.88 | 25.1 | 100 | — | |
| LLaMA-2-7B2024.06 | 5.86 | — | — | — | |
| PiSSA#Params=320M2024.06 | 5.86 | 14.3 | — | — |