Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multi-Hop Search-augmented Question Answering on HotpotQA (success rate (%))
Loading...
34.2
Success Rate
+ MASA
24.424
26.962
29.5
32.038
May 29, 2026
Success Rate
Updated 2d ago
Evaluation Results
Method
Method
Links
Success Rate
+ MASA
Backbone=Qwen3-32B
2026.05
34.2
+ DS-Adapter
Backbone=Qwen3-32B
2026.05
34
+ Base Skill
Backbone=Qwen3-32B
2026.05
33.8
+ MASA
Backbone=Qwen3-14B
2026.05
32.8
+ Base Skill
Backbone=Qwen3-14B
2026.05
32.7
No Skill
Backbone=Qwen3-32B
2026.05
32.2
No Skill
Backbone=Qwen3-14B
2026.05
31.7
+ DS-Adapter
Backbone=Qwen3-14B
2026.05
31.6
+ DS-Adapter
Backbone=Qwen3-4B
2026.05
28.6
+ Base Skill
Backbone=Qwen3-8B
2026.05
28.6
+ MASA
Backbone=Qwen3-8B
2026.05
28.6
+ Base Skill
Backbone=Qwen3-4B
2026.05
28.5
+ DS-Adapter
Backbone=Qwen3-8B
2026.05
27.8
No Skill
Backbone=Qwen3-4B
2026.05
27.7
+ MASA
Backbone=Qwen3-4B
2026.05
27.4
No Skill
Backbone=Qwen3-8B
2026.05
24.8
Feedback
Search any
task
Search any
task