Share your thoughts, 1 month free Claude Pro on usSee more

General Language Evaluation on MMLU, BoolQ, OpenBookQA, RTE

70.4Average Accuracy

Mixtral-8x22B

Updated 9d ago

Evaluation Results

Method	Links
Mixtral-8x22B 2024.07		70.4
Original 2026.04		70.4
GRAPE 2026.04		68.2
Model-centric 2024.07		67.6
Data-centric 2024.07		67
DEK 2026.04		67
Enumerate 2024.07		66.2
Enumerate 2026.04		66.2
Mixtral-8x7B 2024.07		66
Count-guided 2024.07		63.3
Count-guided 2026.04		63.3
Model-centric 2024.07		62.6
GRAPE 2026.04		62.6
Data-centric 2024.07		62.4
Model-centric 2024.07		61.3
Data-centric 2024.07		61.1
DEK 2026.04		61.1
Enumerate 2024.07		59.9
Enumerate 2024.07		59.6
Enumerate 2026.04		59.6
Router-guided 2024.07		57.8
Data-centric 2024.07		54.5
Model-centric 2024.07		54.2
Count-guided 2024.07		54.1
Enumerate 2024.07		50.8
Count-guided 2024.07		50
Count-guided 2026.04		50
Original 2026.04		50
GRAPE 2026.04		49.8
GRAPE 2026.04		49.5
Count-guided 2026.04		48.9
DEK 2026.04		48.8
DEK 2026.04		48.8
Count-guided 2026.04		47.5
Router-guided 2026.04		47.3
Count-guided 2024.07		45.7
Router-guided 2026.04		45.2
Router-guided 2024.07		33.5
Router-guided 2026.04		33.5
Router-guided 2024.07		31.6
Router-guided 2026.04		31.6
Router-guided 2024.07		29.1