Question Answering on CoQA (F1)
76F1 ScoreICD
Evaluation Results
| Method | Links | |
|---|---|---|
| ICDModel=Llama-3.1-8B, IGR (%)=41.4±5.62026.05 | 76 | |
| PC-LDCDModel=Mistral-7B, IGR (%)=58.3±6.02026.05 | 73 | |
| DoLaModel=Llama-3.1-8B, IGR (%)=41.4±5.62026.05 | 72 | |
| ICDModel=Llama-3.2-1B, IGR (%)=77.9±15.42026.05 | 71 | |
| DoLaModel=Mistral-7B, IGR (%)=58.3±6.02026.05 | 71 | |
| SADIModel=Mistral-7B, IGR (%)=58.3±6.02026.05 | 71 | |
| SADIModel=Llama-3.1-8B, IGR (%)=41.4±5.62026.05 | 71 | |
| AdaSteerModel=Llama-3.2-1B, IGR (%)=77.9±15.42026.05 | 70 | |
| SADIModel=Llama-3.2-1B, IGR (%)=77.9±15.42026.05 | 70 | |
| AdaSteerModel=Mistral-7B, IGR (%)=58.3±6.02026.05 | 70 | |
| AdaSteerModel=Llama-3.1-8B, IGR (%)=41.4±5.62026.05 | 70 | |
| PC-LDCDModel=Llama-3.2-1B, IGR (%)=77.9±15.42026.05 | 69 | |
| PC-LDCDModel=Llama-3.1-8B, IGR (%)=41.4±5.62026.05 | 69 | |
| DoLaModel=Llama-3.2-1B, IGR (%)=77.9±15.42026.05 | 68 | |
| ICDModel=Mistral-7B, IGR (%)=58.3±6.02026.05 | 68 | |
| AdaSteerModel=Qwen3-4B, IGR (%)=45.4±3.92026.05 | 58 | |
| SADIModel=Qwen3-4B, IGR (%)=45.4±3.92026.05 | 58 | |
| ICDModel=Qwen3-4B, IGR (%)=45.4±3.92026.05 | 58 | |
| PC-LDCDModel=Qwen3-4B, IGR (%)=45.4±3.92026.05 | 58 | |
| DoLaModel=Qwen3-4B, IGR (%)=45.4±3.92026.05 | 56 | |
| ITIModel=Qwen3-4B, IGR (%)=45.4±3.92026.05 | 53 | |
| ITIModel=Mistral-7B, IGR (%)=58.3±6.02026.05 | 53 | |
| VanillaModel=Llama-3.2-1B, IGR (%)=0.02026.05 | 50 | |
| VanillaModel=Qwen3-4B, IGR (%)=0.02026.05 | 50 | |
| VanillaModel=Mistral-7B, IGR (%)=0.02026.05 | 50 | |
| VanillaModel=Llama-3.1-8B, IGR (%)=0.02026.05 | 50 | |
| ITIModel=Llama-3.1-8B, IGR (%)=41.4±5.62026.05 | 50 | |
| ITIModel=Llama-3.2-1B, IGR (%)=77.9±15.42026.05 | 48 |