Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Multimodal Question Answering on ScienceQA-IMG zero-shot
Loading...
70.4
Accuracy
InstructBLIP
45.544
51.997
58.45
64.903
Dec 9, 2025
Accuracy
Updated 3d ago
Evaluation Results
Method
Method
Links
Accuracy
InstructBLIP
Backbone=FlanT5xL
2025.12
70.4
Qwen2.5-VL
Parameter Scale=72B
2025.12
70.31
Qwen2-VL
Parameter Scale=72B
2025.12
68.12
Qwen2-VL
Parameter Scale=7B
2025.12
63.51
InstructBLIP
Backbone=Vicuna-13B
2025.12
63.12
BLIP-2
Backbone=Vicuna-13B
2025.12
61.01
InstructBLIP
Backbone=Vicuna-7B
2025.12
60.52
Qwen2.5 VL-7B + PHM
Compression Method=PHM...
2025.12
60.5
Qwen2.5-VL
Parameter Scale=3B
2025.12
59.22
LLaVA-1.5-7B + PHM
Compression Method=PHM...
2025.12
55.2
BLIP-2
Backbone=Vicuna-7B
2025.12
53.81
InstructBLIP + PHM
Compression Method=PHM...
2025.12
46.5
Feedback
Search any
task
Search any
task