Granite-3.3-8B-Base is a decoder-only language model with a 128K token context window. It improves upon Granite-3.1-8B-Base by adding support for Fill-in-the-Middle (FIM) using specialized tokens, enabling the model to generate content conditioned on both prefix and suffix. This makes it well-suited for code completion tasks
发布日期2025年4月16日
参数规模8.2B
上下文长度—
许可证Apache 2.0
知识截止2024年4月1日
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| HumanEval | reasoning code | 89.7 | 来源 |
| AttaQ | safety | 88.5 | 来源 |
| HumanEval+ | reasoning | 86.1 | 来源 |
| AIME 2024 | math reasoning | 81.2 | 来源 |
| HellaSwag | reasoning | 80.1 | 来源 |
| TriviaQA | general reasoning | 78.2 | 来源 |
| IFEval | general | 74.8 | 来源 |
| Winogrande | reasoning language | 74.4 | 来源 |
| BIG-Bench Hard | reasoning math language | 69.1 | 来源 |
| MATH-500 | math reasoning | 69.0 | 来源 |
| MMLU | general reasoning language math | 63.9 | 来源 |
| AlpacaEval 2.0 | general creativity reasoning | 62.7 | 来源 |
| GSM8k | math reasoning | 59.0 | 来源 |
| Arena Hard | general reasoning creativity | 57.6 | 来源 |
| TruthfulQA | general reasoning legal healthcare finance | 52.2 | 来源 |
| ARC-C | reasoning general | 50.8 | 来源 |
| AGIEval | reasoning general math | 49.3 | 来源 |
| NQ | reasoning general | 36.5 | 来源 |
| DROP | reasoning math | 36.1 | 来源 |
| PopQA | general reasoning | 26.2 | 来源 |
Pricing
API 价格对比
暂无 API 价格。