Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Video Scene Identification on VSI-Bench
Loading...
38.2
Accuracy
APPO
27.384
30.192
33
35.808
Feb 27, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
APPO
Backbone=Qwen2.5-VL-3B...
2026.02
38.2
APPO
Backbone=Qwen2.5-VL-7B...
2026.02
36.9
DAPO
Backbone=Qwen2.5-VL-3B...
2026.02
36.7
DAPO
Backbone=Qwen2.5-VL-7B...
2026.02
36.6
GRPO
Backbone=Qwen2.5-VL-7B...
2026.02
35.9
GRPO
Backbone=Qwen2.5-VL-3B...
2026.02
34.8
SFT
Backbone=Qwen2.5-VL-3B...
2026.02
32.9
SFT
Backbone=Qwen2.5-VL-7B...
2026.02
31.8
Base Model
Backbone=Qwen2.5-VL-3B...
2026.02
29.2
Base Model
Backbone=Qwen2.5-VL-7B...
2026.02
27.8
Feedback
Search any
task
Search any
task