Qwen2 7B Instruct
Qwen2-7B-Instruct is an instruction-tuned language model with 7 billion parameters, supporting a context length of up to 131,072 tokens.
Benchmark results
| Benchmark | Score | Tags | Source |
|---|---|---|---|
| AlignBench | 72.1% | self-reported llm-stats | link → |
| C-Eval | 77.2% | self-reported llm-stats | link → |
| EvalPlus | 70.3% | self-reported llm-stats | link → |
| GPQA | 25.3% | self-reported llm-stats | link → |
| GSM8k | 82.3% | self-reported llm-stats | link → |
| HumanEval | 79.9% | self-reported llm-stats | link → |
| LiveCodeBench | 26.6% | self-reported llm-stats | link → |
| MATH | 49.6% | self-reported llm-stats | link → |
| MBPP | 67.2% | self-reported llm-stats | link → |
| MMLU | 70.5% | self-reported llm-stats | link → |
| MMLU-Pro | 44.1% | self-reported llm-stats | link → |
| MT-Bench | 84.1 | self-reported llm-stats | link → |
| MultiPL-E | 59.1% | self-reported llm-stats | link → |
| TheoremQA | 25.3% | self-reported llm-stats | link → |