Gemma 3n is a generative AI model optimized for use in everyday devices, such as phones, laptops, and tablets. It features innovations like Per-Layer Embedding (PLE) parameter caching and a MatFormer model architecture for reduced compute and memory. These models handle audio, text, and visual data, though this E4B preview currently supports text and vision input. Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models, and is licensed for responsible commercial use.
发布日期2025年5月20日
参数规模1.9B
上下文长度—
许可证Gemma
知识截止2024年6月1日
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| ARC-E | reasoning general | 81.6 | 来源 |
| BoolQ | language reasoning | 81.6 | 来源 |
| PIQA | reasoning physics general | 81.0 | 来源 |
| HellaSwag | reasoning | 78.6 | 来源 |
| HumanEval | reasoning code | 75.0 | 来源 |
| Winogrande | reasoning language | 71.7 | 来源 |
| TriviaQA | general reasoning | 70.2 | 来源 |
| MMLU | general reasoning language math | 64.9 | 来源 |
| Global-MMLU-Lite | general language reasoning | 64.5 | 来源 |
| MBPP | reasoning general | 63.6 | 来源 |
| ARC-C | reasoning general | 61.6 | 来源 |
| DROP | reasoning math | 60.8 | 来源 |
| MGSM | math reasoning | 60.7 | 来源 |
| Global-MMLU | general language reasoning | 60.3 | 来源 |
| Include | general | 57.2 | 来源 |
| BIG-Bench Hard | reasoning math language | 52.9 | 来源 |
| MMLU-Pro | language reasoning math general | 50.6 | 来源 |
| WMT24++ | language | 50.1 | 来源 |
| Social IQa | reasoning psychology | 50.0 | 来源 |
| HiddenMath | math reasoning | 37.7 | 来源 |
| LiveCodeBench v5 | reasoning general | 25.7 | 来源 |
| GPQA | reasoning general | 23.7 | 来源 |
| Natural Questions | reasoning general search | 20.9 | 来源 |
| MMLU-ProX | language reasoning math general | 19.9 | 来源 |
| Codegolf v2.2 | code | 16.8 | 来源 |
| LiveCodeBench | reasoning general code | 13.2 | 来源 |
| AIME 2025 | math reasoning | 11.6 | 来源 |
| ECLeKTic | language reasoning | 1.9 | 来源 |
Pricing
API 价格对比
暂无 API 价格。