Alibaba Cloud / Qwen Team

Qwen2 7B Instruct

Zero-eval

by Alibaba Cloud / Qwen Team

+
+
+
+
About

Qwen2 7B was created as an efficient variant in the Qwen2 family, designed to provide capable instruction-following with 7 billion parameters. Built to serve as a practical foundation for applications requiring reliable language understanding, it balances performance with deployment efficiency.

+
+
+
+
Timeline
AnnouncedJul 23, 2024
ReleasedJul 23, 2024
+
+
+
+
License & Family
License
Apache 2.0
Performance Overview
Performance metrics and category breakdown

Overall Performance

14 benchmarks
Average Score
59.5%
Best Score
84.1%
High Performers (80%+)
2
+
+
+
+
All Benchmark Results for Qwen2 7B Instruct
Complete list of benchmark scores with detailed information
MT-Bench
text
0.84
84.1%
Self-reported
GSM8k
text
0.82
82.3%
Self-reported
HumanEval
text
0.80
79.9%
Self-reported
C-Eval
text
0.77
77.2%
Self-reported
AlignBench
text
0.72
72.1%
Self-reported
MMLU
text
0.70
70.5%
Self-reported
EvalPlus
text
0.70
70.3%
Self-reported
MBPP
text
0.67
67.2%
Self-reported
MultiPL-E
text
0.59
59.1%
Self-reported
MATH
text
0.50
49.6%
Self-reported
Showing 1 to 10 of 14 benchmarks
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+