Logical Reasoning on GEM mastermind v1 (test)
78.6Pass@1Off. + On.
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Off. + On.Backbone=Qwen3-8B, Protocol=Multi-Task2026.06 | 78.6 | 43.1 | |
| On-policy Distill.Backbone=Qwen3-8B, Protocol=Multi-Task2026.06 | 78 | 41.8 | |
| RLBackbone=Qwen3-8B, Protocol=Single-Task2026.06 | 77.3 | 38.7 | |
| Off-policy Distill.Backbone=Qwen3-8B, Protocol=Single-Task2026.06 | 77.1 | 41.7 | |
| Off-policy Distill.Backbone=Qwen3-8B, Protocol=Multi-Task2026.06 | 76.5 | 39.6 | |
| Pre-RLBackbone=Qwen3-8B, Protocol=Single-Task2026.06 | 3.6 | 0 |