Applied Mathematics on Graduate-level Applied Mathematics final exam (pilot study)
90AccuracyShadow (Forced Tools)
Evaluation Results
| Method | Links | |
|---|---|---|
| Shadow (Forced Tools)Base Model=Qwen3-32B2026.03 | 90 | |
| Shadow (Full/Dynamic)Base Model=Qwen3-32B2026.03 | 85 | |
| Shadow (No Code)Base Model=Qwen3-32B2026.03 | 85 | |
| Naive RAGBase Model=Qwen3-32B2026.03 | 74 | |
| Baseline (Zero-shot)Base Model=Qwen3-32B2026.03 | 67 | |
| Shadow (Full/Dynamic)Base Model=Qwen2.5-32B2026.03 | 65 | |
| Shadow (Forced Tools)Base Model=Qwen2.5-32B2026.03 | 57 | |
| Naive RAGBase Model=Qwen2.5-32B2026.03 | 56 | |
| Shadow (No Code)Base Model=Qwen2.5-32B2026.03 | 50 | |
| Baseline (Zero-shot)Base Model=Qwen2.5-32B2026.03 | 47 |