Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
OCR-based Visual Question Answering on TextVQA
Loading...
61.2
VQAText
LLaVA-1.5-13B
52.36
54.655
56.95
59.245
Jun 17, 2026
VQAText
Relative Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
VQAText
Relative Accuracy
LLaVA-1.5-13B
Backbone=LLaVA-1.5-13B...
2026.06
61.2
100
SparseVLM
Backbone=LLaVA-1.5-13B...
2026.06
59.1
97.9
DART
Backbone=LLaVA-1.5-13B...
2026.06
59
97.1
SPARE
Backbone=LLaVA-1.5-13B...
2026.06
58.6
98
CDPruner
Backbone=LLaVA-1.5-13B...
2026.06
58.4
97.9
SPARE
Backbone=LLaVA-1.5-13B...
2026.06
58.1
97.2
CDPruner
Backbone=LLaVA-1.5-13B...
2026.06
57.6
96.3
PruMerge+
Backbone=LLaVA-1.5-13B...
2026.06
56.6
93.2
SPARE
Backbone=LLaVA-1.5-13B...
2026.06
56.3
94.7
SparseVLM
Backbone=LLaVA-1.5-13B...
2026.06
55.8
89.7
DART
Backbone=LLaVA-1.5-13B...
2026.06
55.8
92.5
PruMerge+
Backbone=LLaVA-1.5-13B...
2026.06
55.7
90
CDPruner
Backbone=LLaVA-1.5-13B...
2026.06
55.2
92.6
DART
Backbone=LLaVA-1.5-13B...
2026.06
52.7
86
Feedback
Search any
task
Search any
task