Mathematical Reasoning on MATH-500 (Score, Retention)
96.4ScoreHaiku 4.5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Haiku 4.5Training Phase=Original (Orig)2026.03 | 96.4 | — | |
| Haiku 4.5Training Phase=Trojan-Speak (TS)2026.03 | 93.3 | 97 | |
| Qwen3 14BTraining Phase=Original (Orig)2026.03 | 87.1 | — | |
| Qwen3 32BTraining Phase=Original (Orig)2026.03 | 86.9 | — | |
| Qwen3 4BTraining Phase=Original (Orig)2026.03 | 84.3 | — | |
| Qwen3 8BTraining Phase=Original (Orig)2026.03 | 82.8 | — | |
| Qwen3 32BTraining Phase=Trojan-Speak (TS)2026.03 | 82.4 | 95 | |
| Qwen3 14BTraining Phase=Trojan-Speak (TS)2026.03 | 79.4 | 92 | |
| Qwen3 8BTraining Phase=Trojan-Speak (TS)2026.03 | 64.3 | 78 | |
| Qwen3 4BTraining Phase=Trojan-Speak (TS)2026.03 | 58.7 | 70 |