General Knowledge on MMLU-Pro (pass@1)
66.46pass@1Qwen3-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-8BBackbone=Qwen3-8B2026.01 | 66.46 | |
| RelayLLM (Difficulty-Aware)Student Model Backbone=Qwen3-1.7B2026.01 | 59.03 | |
| RelayLLM (Simple)Student Model Backbone=Qwen3-1.7B2026.01 | 58.76 | |
| SRFT [13]Base Model=Qwen2.5-Math-7B, Source=reported2026.04 | 55.9 | |
| SFT→RLBase Model=Qwen2.5-Math-7B, Source=verl2026.04 | 55.1 | |
| ReLIFT [12]Base Model=Qwen2.5-Math-7B, Source=reported2026.04 | 53.9 | |
| CITERStudent Model Backbone=Qwen3-1.7B2026.01 | 53.38 | |
| LUFFY [11]Base Model=Qwen2.5-Math-7B, Source=reported2026.04 | 53 | |
| HPT [15]Base Model=Qwen2.5-Math-7B, Source=reported2026.04 | 52.5 | |
| Prefix-RFT [14]Base Model=Qwen2.5-Math-7B, Source=reported2026.04 | 52.1 | |
| ReLIFT [12]Base Model=Qwen2.5-Math-7B, Source=reproduction attempt2026.04 | 51.3 | |
| LUFFY [11]Base Model=Qwen2.5-Math-7B, Source=reproduction attempt2026.04 | 50.9 | |
| SFTBase Model=Qwen2.5-Math-7B, Source=verl2026.04 | 50.2 | |
| GRPOStudent Model Backbone=Qwen3-1.7B2026.01 | 49.76 | |
| Base ModelStudent Model Backbone=Qwen3-1.7B2026.01 | 46.9 | |
| RelayLLM (Difficulty-Aware)Student Model Backbone=Qwen3-0.6B2026.01 | 35.87 | |
| RelayLLM (Simple)Student Model Backbone=Qwen3-0.6B2026.01 | 35.61 | |
| CITERStudent Model Backbone=Qwen3-0.6B2026.01 | 33.12 | |
| GRPOStudent Model Backbone=Qwen3-0.6B2026.01 | 32.15 | |
| Base ModelStudent Model Backbone=Qwen3-0.6B2026.01 | 30.03 |