Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Compositional Reasoning on NaturalBench
Loading...
35.5
Accuracy
InternVL3.5-14B +FINER-Tuning
14.908
20.254
25.6
30.946
Mar 18, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
InternVL3.5-14B +FINER-Tuning
Backbone=InternVL3.5-1...
2026.03
35.5
Qwen2.5-VL-7B
Backbone=Qwen2.5-VL-7B
2026.03
34.1
Qwen2.5-VL-7B +FINER-Tuning
Backbone=Qwen2.5-VL-7B...
2026.03
34.1
InternVL3.5-8B +FINER-Tuning
Backbone=InternVL3.5-8...
2026.03
31.1
InternVL3.5-14B
Backbone=InternVL3.5-14B
2026.03
30.7
InternVL3.5-8B
Backbone=InternVL3.5-8B
2026.03
30.4
OmniLMM-12B
Backbone=OmniLMM-12B
2026.03
26.9
LLaVA-1.6-7B +FINER-Tuning
Backbone=LLaVA-1.6-7B,...
2026.03
19.8
OmniLMM-12B +RLAIF-V
Backbone=OmniLMM-12B,...
2026.03
19.4
LLaVA-1.6-7B
Backbone=LLaVA-1.6-7B
2026.03
15.7
Feedback
Search any
task
Search any
task