Natural Language Inference on WinoGrande
53.49AccuracyCOLLATE
Evaluation Results
| Method | Links | |
|---|---|---|
| COLLATEBackbone=open-llama-v2-7b2025.06 | 53.49 | |
| Distilling Step-by-StepCategory=Task-specific Supervised Fine-Tuning, Backbone=open-llama-v2-7b2025.06 | 50.69 | |
| SPINCategory=Training-driven Rationale Enhancement, Backbone=open-llama-v2-7b2025.06 | 50.58 | |
| open-llama-7b-v2 SFTCategory=Task-specific Supervised Fine-Tuning, Backbone=open-llama-v2-7b2025.06 | 48.81 | |
| Self-Rewarding LMsCategory=Training-driven Rationale Enhancement, Backbone=open-llama-v2-7b2025.06 | 41.44 | |
| CoT Self-ConsistencyCategory=Prompt-based Rationale Refinement, Backbone=open-llama-v2-7b2025.06 | 33.32 | |
| Exchange-of-ThoughtCategory=Prompt-based Rationale Refinement, Backbone=open-llama-v2-7b2025.06 | 33.29 | |
| Tree-of-ThoughtCategory=Prompt-based Rationale Refinement, Backbone=open-llama-v2-7b2025.06 | 30.1 | |
| Chain-of-ThoughtCategory=Prompt-based Rationale Refinement, Backbone=open-llama-v2-7b2025.06 | 28.19 |