Arithmetic Reasoning on Arithmetic (test)
93.2Avg@16RLVR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RLVRZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT2025.10 | 93.2 | 95.5 | |
| RLERZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT, k=4, Gk=82025.10 | 71.5 | 75.8 | |
| RLERZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT, k=2, Gk=82025.10 | 69.2 | 72.2 | |
| RLIR INTUITORZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT2025.10 | 60.4 | 61.5 | |
| RLIR SCZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT2025.10 | 57.4 | 60.2 | |
| RLIR FreqZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT2025.10 | 56.9 | 62.6 | |
| RLIR JudgeZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT2025.10 | 48.3 | 70.6 | |
| pre-RLZero-shot=true, Model Backbone=QWEN-2.5-1.5B-INSTRUCT2025.10 | 41.5 | 89.2 |