Logic Reasoning on ARC-Challenge & LogiQA OpenCompass (test)
38.31ARC-C AccuracyCRITIQ
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CRITIQBackbone=Llama-3.2-3B, Continual Pretraining=true, Data Selection Strategy=CRITIQ Scorer2025.02 | 38.31 | 30.41 | 34.36 | |
| RawBackbone=Llama-3.2-3B, Continual Pretraining=false2025.02 | 37.97 | 27.34 | 32.66 | |
| Zyda-2Backbone=Llama-3.2-3B, Continual Pretraining=true, Data Selection Strategy=Zyda-2 subset2025.02 | 36.61 | 23.5 | 30.06 | |
| QR-EduBackbone=Llama-3.2-3B, Continual Pretraining=true, Data Selection Strategy=QuRators (Educational Value)2025.02 | 35.59 | 26.88 | 31.24 |