Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multi-hop QA on 2WikiMHQA Wiki18 index (test)
Loading...
55.1
Pass@4
Qwen3-8B
23.796
31.923
40.05
48.177
Jun 15, 2026
Pass@4
Updated 1mo ago
Evaluation Results
Method
Method
Links
Pass@4
Qwen3-8B
Sampling Method=DivIni...
2026.06
55.1
Gemma3-12B
Sampling Method=DivIni...
2026.06
53.9
Gemma3-4B
Sampling Method=DivIni...
2026.06
52.2
Gemma3-12B
Sampling Method=Standa...
2026.06
52
Gemma3-12B
Sampling Method=DivIni...
2026.06
50.2
Qwen3-4B
Sampling Method=DivIni...
2026.06
49
Qwen3-8B
Sampling Method=Standa...
2026.06
46.3
Qwen3-8B
Sampling Method=DivIni...
2026.06
46
Gemma3-12B
Sampling Method=Standa...
2026.06
45
Gemma3-4B
Sampling Method=Standa...
2026.06
42.8
Qwen3-4B
Sampling Method=Standa...
2026.06
41.9
Qwen3-1.7B
Sampling Method=DivIni...
2026.06
41.5
Qwen3-8B
Sampling Method=Standa...
2026.06
38.6
Qwen3-1.7B
Sampling Method=Standa...
2026.06
37.6
Qwen3-4B
Sampling Method=DivIni...
2026.06
36.6
Gemma3-4B
Sampling Method=DivIni...
2026.06
34
Qwen3-4B
Sampling Method=Standa...
2026.06
29.5
Gemma3-4B
Sampling Method=Standa...
2026.06
28
Qwen3-1.7B
Sampling Method=DivIni...
2026.06
27.8
Qwen3-1.7B
Sampling Method=Standa...
2026.06
25
Feedback
Search any
task
Search any
task