Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Overall Capability Merging on Multi-Task Average
Loading...
56.56
Avg Accuracy
SWUDI
48.0008
50.2229
52.445
54.6671
Jun 5, 2026
Avg Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Avg Accuracy
SWUDI
mode=Full fine-tuning,...
2026.06
56.56
SWUDI-A
mode=Full fine-tuning,...
2026.06
56.33
Mixture Training
mode=Full fine-tuning,...
2026.06
56.22
OptMerge
mode=Full fine-tuning,...
2026.06
56.18
WUDI Merging
mode=Full fine-tuning,...
2026.06
56
TIES Merging
mode=Full fine-tuning,...
2026.06
55.87
TIES w/ DARE
mode=Full fine-tuning,...
2026.06
55.85
Task Arithmetic
mode=Full fine-tuning,...
2026.06
55.35
TA w/ DARE
mode=Full fine-tuning,...
2026.06
55.17
Iso-C
mode=Full fine-tuning,...
2026.06
53.8
InternVL2.5-Instruct
mode=Full fine-tuning,...
2026.06
53.48
TSV Merging
mode=Full fine-tuning,...
2026.06
52.5
Weight Average
mode=Full fine-tuning,...
2026.06
48.33
Feedback
Search any
task
Search any
task