Mathematical Reasoning on AMC 22 23 (Mean@4)
83.4Mean@4Sparsity setup (0.92) Rollout
Evaluation results
| Method | Links | |
|---|---|---|
| Sparsity setup (0.92) RolloutModel Backbone=Qwen3-1.7B, Sparsity Level=0.922026.06 | 83.4 | |
| Sparsity setup (0.86) RolloutModel Backbone=Qwen3-8B, Sparsity Level=0.862026.06 | 82.77 | |
| Sparsity setup (0.86) RolloutModel Backbone=Qwen3-1.7B, Sparsity Level=0.862026.06 | 81.92 | |
| Dense Rollout Dense TrainingModel Backbone=Qwen3-8B2026.06 | 81.11 | |
| Sparsity setup (0.92) RolloutModel Backbone=Qwen3-8B, Sparsity Level=0.922026.06 | 81.11 | |
| Sparsity setup (0.86) RolloutModel Backbone=Qwen3-14B, Sparsity Level=0.862026.06 | 81.11 | |
| Dense Rollout Dense TrainingModel Backbone=Qwen3-1.7B2026.06 | 80.72 | |
| Dense Rollout Dense TrainingModel Backbone=Qwen3-14B2026.06 | 80.55 | |
| Dense Rollout Dense TrainingModel Backbone=Qwen3-4B2026.06 | 80 | |
| Sparsity setup (0.92) RolloutModel Backbone=Qwen3-4B, Sparsity Level=0.922026.06 | 79.44 | |
| Sparsity setup (0.86) RolloutModel Backbone=Qwen3-4B, Sparsity Level=0.862026.06 | 79.44 | |
| Sparse with LoRA Distillation (DistillSparse)Model Backbone=Qwen3-1.7B2026.06 | 73.33 |