Instruction Following on ComplexInstruct Level 3
88.7ISRDVRws
Evaluation Results
| Method | Links | |
|---|---|---|
| DVRwsBackbone=Llama-3.1-70B-Instruct-AWQ-INT4, warm-start=true2024.10 | 88.7 | |
| DVRcsBackbone=Llama-3.1-70B-Instruct-AWQ-INT4, cold-start=true2024.10 | 88.2 | |
| CRITICBackbone=Llama-3.1-70B-Instruct-AWQ-INT42024.10 | 87.6 | |
| ReActBackbone=Llama-3.1-70B-Instruct-AWQ-INT42024.10 | 83.5 | |
| R-SampleBackbone=Llama-3.1-70B-Instruct-AWQ-INT42024.10 | 83.2 | |
| DVRBackbone=Llama3.1-8B, Evaluation Protocol=zero-shot2024.10 | 79.2 | |
| GPT-4-turboEvaluation Protocol=zero-shot2024.10 | 78.8 | |
| DVRBackbone=Llama-3-8B-Instruct, Start Type=Warm-Start2024.10 | 77 | |
| DVRBackbone=Llama-3-8B-Instruct, Start Type=Cold-Start2024.10 | 76.9 | |
| CRITICBackbone=Llama-3-8B-Instruct2024.10 | 74.4 | |
| ReflexionBackbone=Llama-3.1-70B-Instruct-AWQ-INT42024.10 | 72.8 | |
| BSMBackbone=Llama-3.1-70B-Instruct-AWQ-INT42024.10 | 72.6 | |
| VanillaBackbone=Llama-3.1-70B-Instruct-AWQ-INT42024.10 | 72.4 | |
| U-SCBackbone=Llama-3.1-70B-Instruct-AWQ-INT42024.10 | 71.8 | |
| ReActBackbone=Llama-3-8B-Instruct2024.10 | 68.8 | |
| DVRBackbone=Mistral-7B, Evaluation Protocol=zero-shot2024.10 | 66.6 | |
| R-SampleBackbone=Llama-3-8B-Instruct2024.10 | 64.5 | |
| Llama3.1-8BEvaluation Protocol=zero-shot2024.10 | 62.5 | |
| ReflexionBackbone=Llama-3-8B-Instruct2024.10 | 57.5 | |
| U-SCBackbone=Llama-3-8B-Instruct2024.10 | 56.7 | |
| VanillaBackbone=Llama-3-8B-Instruct2024.10 | 56.4 | |
| BSMBackbone=Llama-3-8B-Instruct2024.10 | 56 | |
| Mistral-7BEvaluation Protocol=zero-shot2024.10 | 34.1 |