language: zh pipeline_tag: zero-shot-classification tags:

• transformers datasets:
• nyu-mll/multi_nli
• stanfordnlp/snli metrics:
• accuracy license: apache-2.0 base_model:
• microsoft/deberta-base library_name: sentence-transformers

自然语言推理交叉编码器

本模型采用SentenceTransformers框架的交叉编码器进行训练。

训练数据

模型基于SNLI和MultiNLI数据集训练。对于给定的句子对，模型将输出三个分数，分别对应矛盾、蕴含和中立三种标签。

性能表现

评估结果详见SBERT.net - 预训练交叉编码器。

使用方法

预训练模型调用方式如下：

from sentence_transformers import CrossEncoder
model = CrossEncoder('cross-encoder/nli-deberta-base')
scores = model.predict([('一个男人正在吃披萨', '一个男人在吃东西'), ('一辆黑色赛车在人群前启动', '一个男人独自驾车行驶在荒凉的公路上')])

#将分数转换为标签
label_mapping = ['矛盾', '蕴含', '中立']
labels = [label_mapping[score_max] for score_max in scores.argmax(axis=1)]

Transformers库直接调用

也可不依赖SentenceTransformers库，直接使用Transformers库调用：

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model = AutoModelForSequenceClassification.from_pretrained('cross-encoder/nli-deberta-base')
tokenizer = AutoTokenizer.from_pretrained('cross-encoder/nli-deberta-base')

features = tokenizer(['一个男人正在吃披萨', '一辆黑色赛车在人群前启动'], ['一个男人在吃东西', '一个男人独自驾车行驶在荒凉的公路上'],  padding=True, truncation=True, return_tensors="pt")

model.eval()
with torch.no_grad():
    scores = model(**features).logits
    label_mapping = ['矛盾', '蕴含', '中立']
    labels = [label_mapping[score_max] for score_max in scores.argmax(dim=1)]
    print(labels)

零样本分类

该模型还可用于零样本分类任务：

from transformers import pipeline

classifier = pipeline("zero-shot-classification", model='cross-encoder/nli-deberta-base')

sent = "苹果公司刚刚发布了最新款iPhone X"
candidate_labels = ["科技", "体育", "政治"]
res = classifier(sent, candidate_labels)
print(res)