Question Answering on SealQA 111 questions
49.5AccuracySkillSmith
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SkillSmithBackbone=Qwen3.5-397B2026.05 | 49.5 | 55 | |
| EvoSkillBackbone=Qwen3.5-397B2026.05 | 41.4 | 46 | |
| Base AgentBackbone=Qwen3.5-397B2026.05 | 30.6 | 34 | |
| SkillSmithBackbone=Qwen3.5-122B2026.05 | 27.9 | 31 | |
| EvoSkillBackbone=Qwen3.5-122B2026.05 | 23.4 | 26 | |
| Base AgentBackbone=Qwen3.5-122B2026.05 | 17.1 | 19 | |
| SkillSmithBackbone=Qwen3.5-35B2026.05 | 17.1 | 19 | |
| EvoSkillBackbone=Qwen3.5-35B2026.05 | 14.4 | 16 | |
| Base AgentBackbone=Qwen3.5-35B2026.05 | 10.8 | 12 | |
| SkillSmithBackbone=Qwen3.5-9B2026.05 | 6.3 | 7 | |
| EvoSkillBackbone=Qwen3.5-9B2026.05 | 5.4 | 6 | |
| Base AgentBackbone=Qwen3.5-9B2026.05 | 4.5 | 5 |