| | --- |
| | license: mit |
| | library_name: transformers |
| | tags: |
| | - pytorch |
| | - text-classification |
| | - sentiment-analysis |
| | - chinese |
| | pipeline_tag: text-classification |
| | language: zh |
| | --- |
| | |
| | # Tiny Sentiment Classifier |
| |
|
| | 这是一个非常小的自定义 Transformer 模型,用于**中文情感二分类**(正面/负面)。 |
| |
|
| | ## 模型简介 |
| |
|
| | - 架构:小型 Transformer Encoder + 线性分类头 |
| | - 隐藏维度:128 |
| | - 层数:2 |
| | - 词汇表:字符级(支持中文汉字 + ASCII) |
| | - 训练数据:极简玩具数据集(仅用于演示 HF 适配流程) |
| |
|
| | ## 使用方式 |
| |
|
| | ```python |
| | from transformers import pipeline |
| | |
| | # 加载模型 |
| | classifier = pipeline( |
| | "text-classification", |
| | model="huiqian/tiny-sentiment-classifier" |
| | ) |
| | |
| | # 预测 |
| | text = "这家店的服务超级好,强烈推荐!" |
| | result = classifier(text) |
| | print(result) |
| | # 示例输出: [{'label': '正面', 'score': 0.98}] |