Constraint-Following on ComplexInstruct & CoDI (Level 4)
69.7ISRDVR_WS
Evaluation Results
| Method | Links | |
|---|---|---|
| DVR_WSBackbone=Llama-3.1-8B-Instruct, Refinement Repository=warm-start2024.10 | 69.7 | |
| DVR_CSBackbone=Llama-3.1-8B-Instruct, Refinement Repository=cold-start2024.10 | 69.5 | |
| CRITICBackbone=Llama-3.1-8B-Instruct2024.10 | 64.4 | |
| ReActBackbone=Llama-3.1-8B-Instruct2024.10 | 60.7 | |
| R-SampleBackbone=Llama-3.1-8B-Instruct2024.10 | 60.4 | |
| VanillaBackbone=Llama-3.1-8B-Instruct2024.10 | 50.1 | |
| ReflxionBackbone=Llama-3.1-8B-Instruct2024.10 | 49.8 | |
| BSMBackbone=Llama-3.1-8B-Instruct2024.10 | 47.5 | |
| U-SCBackbone=Llama-3.1-8B-Instruct2024.10 | 47.1 |