Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Logit Divergence Analysis on Physical reasoning domain (calibration set)
Loading...
0.92
Mean KL Divergence (nats)
µCRASP
0.5364
3.1257
5.715
8.3043
May 25, 2026
Mean KL Divergence (nats)
Ratio vs. µCRASP
Updated 8d ago
Evaluation Results
Method
Method
Links
Mean KL Divergence (nats)
Ratio vs. µCRASP
µCRASP
Base Model=Qwen2.5-VL-...
2026.05
0.92
1
LLM-Pruner
Base Model=Qwen2.5-VL-...
2026.05
1.53
1.7
Wanda-SP
Base Model=Qwen2.5-VL-...
2026.05
3.31
3.6
FLAP
Base Model=Qwen2.5-VL-...
2026.05
4.11
4.5
Attribution Pruning
Base Model=Qwen2.5-VL-...
2026.05
10.51
11.4
Feedback
Search any
task
Search any
task