Gemma 2 9B IT is an instruction-tuned version of Google's Gemma 2 9B base model. It was trained on 8 trillion tokens of web data, code, and math content. The model features sliding window attention, logit soft-capping, and knowledge distillation techniques. It's optimized for dialogue applications through supervised fine-tuning, distillation, RLHF, and model merging using WARP.
发布日期2024年6月27日
参数规模9.2B
上下文长度—
许可证Gemma
知识截止—
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| ARC-E | reasoning general | 88.0 | 来源 |
| BoolQ | language reasoning | 84.2 | 来源 |
| HellaSwag | reasoning | 81.9 | 来源 |
| PIQA | reasoning physics general | 81.7 | 来源 |
| Winogrande | reasoning language | 80.6 | 来源 |
| TriviaQA | general reasoning | 76.6 | 来源 |
| MMLU | general reasoning language math | 71.3 | 来源 |
| GSM8k | math reasoning | 68.6 | 来源 |
| ARC-C | reasoning general | 68.4 | 来源 |
| BIG-Bench | reasoning math language | 68.2 | 来源 |
| Social IQa | reasoning psychology | 53.4 | 来源 |
| AGIEval | reasoning general math | 52.8 | 来源 |
| MBPP | reasoning general | 52.4 | 来源 |
| HumanEval | reasoning code | 40.2 | 来源 |
| MATH | math reasoning | 36.6 | 来源 |
| Natural Questions | reasoning general search | 29.2 | 来源 |
Pricing
API 价格对比
暂无 API 价格。