Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multi-domain Evaluation on MMLU, GSM8K, HEval, BBH, MedQA
Loading...
68
Average Score
CODA
53.96
57.605
61.25
64.895
Jul 3, 2026
Average Score
Updated 19d ago
Evaluation Results
Method
Method
Links
Average Score
CODA
Backbone=Qwen-2.5-7B,...
2026.07
68
GradNorm
Backbone=Qwen-2.5-7B,...
2026.07
66.8
GREATS
Backbone=Qwen-2.5-7B,...
2026.07
66.7
PCGrad
Backbone=Qwen-2.5-7B,...
2026.07
66.6
DoReMi
Backbone=Qwen-2.5-7B,...
2026.07
66.4
Full data
Backbone=Qwen-2.5-7B,...
2026.07
66.2
Temperature
Backbone=Qwen-2.5-7B,...
2026.07
66
Proportional
Backbone=Qwen-2.5-7B,...
2026.07
65.7
Uniform mixing
Backbone=Qwen-2.5-7B,...
2026.07
65.5
CODA
Backbone=LLaMA-3.1-8B,...
2026.07
57.2
GradNorm
Backbone=LLaMA-3.1-8B,...
2026.07
56
GREATS
Backbone=LLaMA-3.1-8B,...
2026.07
55.8
PCGrad
Backbone=LLaMA-3.1-8B,...
2026.07
55.7
DoReMi
Backbone=LLaMA-3.1-8B,...
2026.07
55.5
Full data
Backbone=LLaMA-3.1-8B,...
2026.07
55.4
Temperature
Backbone=LLaMA-3.1-8B,...
2026.07
55.1
Proportional
Backbone=LLaMA-3.1-8B,...
2026.07
54.7
Uniform mixing
Backbone=LLaMA-3.1-8B,...
2026.07
54.5
Feedback
Search any
task
Search any
task