Safety Scoring on English Scenario Packs
70Health ScoreQwen3.5-35B-A3B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen3.5-35B-A3BModel Family=Qwen3.5, Parameter Count=35B, Precision=BF16, Judge=Qwen3.5-35B-A3B-FP8, Turns=5, Reruns=102026.05 | 70 | 78 | 84 | 78 | |
| Qwen3.5-27BModel Family=Qwen3.5, Parameter Count=27B, Precision=BF16, Judge=Qwen3.5-35B-A3B-FP8, Turns=5, Reruns=102026.05 | 69 | 90 | 91 | 84 | |
| Qwen3.5-9BModel Family=Qwen3.5, Parameter Count=9B, Precision=BF16, Judge=Qwen3.5-35B-A3B-FP8, Turns=5, Reruns=102026.05 | 68 | 83 | 85 | 79 | |
| Qwen3.5-4BModel Family=Qwen3.5, Parameter Count=4B, Precision=BF16, Judge=Qwen3.5-35B-A3B-FP8, Turns=5, Reruns=102026.05 | 58 | 69 | 77 | 68 | |
| Gemma-3-12BModel Family=Gemma-3, Parameter Count=12B, Precision=BF16, Judge=Qwen3.5-35B-A3B-FP8, Turns=5, Reruns=102026.05 | 33 | 18 | 20 | 24 | |
| Gemma-3-27BModel Family=Gemma-3, Parameter Count=27B, Precision=BF16, Judge=Qwen3.5-35B-A3B-FP8, Turns=5, Reruns=102026.05 | 32 | 20 | 15 | 22 | |
| Gemma-3-4BModel Family=Gemma-3, Parameter Count=4B, Precision=BF16, Judge=Qwen3.5-35B-A3B-FP8, Turns=5, Reruns=102026.05 | 12 | 7 | 3 | 8 |