General Language Modeling on MMLU, ARC-Challenge, and CommonsenseQA Aggregate
64.77Average ScoreRAISE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RAISEModel=Qwen-2.5-3B2025.04 | 64.77 | 69.83 | |
| SSPLModel=Qwen-2.5-3B2025.04 | 64.42 | 46.59 | |
| IFDModel=Qwen-2.5-3B2025.04 | 64.22 | 34.04 | |
| AlpaGasusModel=Qwen-2.5-3B2025.04 | 64.19 | 32.06 | |
| DEITAModel=Qwen-2.5-3B2025.04 | 64.13 | 28.37 | |
| RANDModel=Qwen-2.5-3B2025.04 | 63.94 | 15.53 | |
| Full Alpaca (100%)Model=Qwen-2.5-3B2025.04 | 63.7 | 0 | |
| Base Model (0%)Model=Qwen-2.5-3B2025.04 | 62.16 | -100 | |
| RAISEModel=Llama-3.2-3B2025.04 | 55.47 | 70.35 | |
| IFDModel=Llama-3.2-3B2025.04 | 54.73 | 25.06 | |
| Full Alpaca (100%)Model=Llama-3.2-3B2025.04 | 54.32 | 0 | |
| RANDModel=Llama-3.2-3B2025.04 | 54.2 | -7.33 | |
| DEITAModel=Llama-3.2-3B2025.04 | 54.05 | -16.44 | |
| AlpaGasusModel=Llama-3.2-3B2025.04 | 53.13 | -72.14 | |
| Base Model (0%)Model=Llama-3.2-3B2025.04 | 52.67 | -100 | |
| SSPLModel=Llama-3.2-3B2025.04 | 51.08 | -196.75 | |
| RAISEModel=Llama-3.2-1B2025.04 | 40.24 | 85.14 | |
| SSPLModel=Llama-3.2-1B2025.04 | 40.08 | 69.55 | |
| RANDModel=Llama-3.2-1B2025.04 | 39.44 | 7.01 | |
| Full Alpaca (100%)Model=Llama-3.2-1B2025.04 | 39.36 | 0 | |
| IFDModel=Llama-3.2-1B2025.04 | 39.2 | -15.77 | |
| AlpaGasusModel=Llama-3.2-1B2025.04 | 38.88 | -47.39 | |
| DEITAModel=Llama-3.2-1B2025.04 | 38.69 | -65.79 | |
| Base Model (0%)Model=Llama-3.2-1B2025.04 | 38.33 | -100 |