language: zh widget:

• text: "我喜欢下雨。"
• text: "我讨厌他。"

liam168/c2-roberta-base-finetuned-dianping-chinese

模型描述

基于中文对话情绪语料训练的文本分类模型，支持2分类：乐观（positive）和悲观（negative）。

概览

• 语言模型: BertForSequenceClassification
• 模型大小: 410M
• 支持语言: 中文

使用示例

>>> from transformers import AutoModelForSequenceClassification , AutoTokenizer, pipeline

>>> model_name = "liam168/c2-roberta-base-finetuned-dianping-chinese"
>>> class_num = 2
>>> ts_texts = ["我喜欢下雨。", "我讨厌他."]
>>> model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=class_num)
>>> tokenizer = AutoTokenizer.from_pretrained(model_name)
>>> classifier = pipeline('sentiment-analysis', model=model, tokenizer=tokenizer)
>>> classifier(ts_texts[0])
>>> classifier(ts_texts[1])

[{'label': 'positive', 'score': 0.9973447918891907}]
[{'label': 'negative', 'score': 0.9972558617591858}]