e5-base-v2
简介
弱监督对比预训练的文本嵌入。 梁旺、杨楠、黄晓龙、焦斌星、杨林军、江大新、Rangan Majumder、韦福如,arXiv 2022
模型卡片
模型配置
模型详情
已翻译E5-base-v2
Text Embeddings by Weakly-Supervised Contrastive Pre-training
Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, Furu Wei, arXiv 2022
该模型包含12层,embedding大小为768。
使用方法
以下是一个示例,用于对MS-MARCO段落排序数据集中的查询和段落进行编码。
import torch.nn.functional as F
from torch import Tensor
from transformers import AutoTokenizer, AutoModel
def average_pool(last_hidden_states: Tensor,
attention_mask: Tensor) -> Tensor:
last_hidden = last_hidden_states.masked_fill(~attention_mask[..., None].bool(), 0.0)
return last_hidden.sum(dim=1) / attention_mask.sum(dim=1)[..., None]
# Each input text should start with "query: " or "passage: ".
# For tasks other than retrieval, you can simply use the "query: " prefix.
input_texts = ['query: how much protein should a female eat',
'query: summit define',
"passage: As a general guideline, the CDC's average requirement of protein for women ages 19 to 70 is 46 grams per day. But, as you can see from this chart, you'll need to increase that if you're expecting or training for a marathon. Check out the chart below to see how much protein you should be eating each day.",
"passage: Definition of summit for English Language Learners. : 1 the highest point of a mountain : the top of a mountain. : 2 the highest level. : 3 a meeting or series of meetings between the leaders of two or more governments."]
tokenizer = AutoTokenizer.from_pretrained('intfloat/e5-base-v2')
model = AutoModel.from_pretrained('intfloat/e5-base-v2')
# Tokenize the input texts
batch_dict = tokenizer(input_texts, max_length=512, padding=True, truncation=True, return_tensors='pt')
outputs = model(**batch_dict)
embeddings = average_pool(outputs.last_hidden_state, batch_dict['attention_mask'])
# normalize embeddings
embeddings = F.normalize(embeddings, p=2, dim=1)
scores = (embeddings[:2] @ embeddings[2:].T) * 100
print(scores.tolist())
训练细节
请参考我们的论文:https://arxiv.org/pdf/2212.03533.pdf
基准评估
请查看 unilm/e5 以复现
BEIR 和 MTEB benchmark 上的评估结果。
对 Sentence Transformers 的支持
以下是与 sentence_transformers 配合使用的示例。
from sentence_transformers import SentenceTransformer
model = SentenceTransformer('intfloat/e5-base-v2')
input_texts = [
'query: how much protein should a female eat',
'query: summit define',
"passage: As a general guideline, the CDC's average requirement of protein for women ages 19 to 70 is 46 grams per day. But, as you can see from this chart, you'll need to increase that if you're expecting or training for a marathon. Check out the chart below to see how much protein you should be eating each day.",
"passage: Definition of summit for English Language Learners. : 1 the highest point of a mountain : the top of a mountain. : 2 the highest level. : 3 a meeting or series of meetings between the leaders of two or more governments."
]
embeddings = model.encode(input_texts, normalize_embeddings=True)
包依赖
pip install sentence_transformers~=2.2.2
贡献者:michaelfeil
常见问题
1. 我需要在输入文本中添加前缀 "query: " 和 "passage: " 吗?
是的,这是模型的训练方式,否则你会看到性能下降。
以下是一些经验法则:
- 对于非对称任务(如开放QA中的段落检索、即席信息检索),相应地使用 "query: " 和 "passage: "。
-
对于对称任务(如语义相似度、释义检索),使用 "query: " 前缀。
-
如果你想将embedding用作特征(如线性探测分类、聚类),使用 "query: " 前缀。
2. 为什么我复现的结果与模型卡片中报告的结果略有不同?
不同版本的 transformers 和 pytorch 可能会导致可忽略但非零的性能差异。
3. 为什么余弦相似度分数分布在0.7到1.0之间?
这是已知且预期的行为,因为我们在InfoNCE对比损失中使用了0.01的低温度系数。
对于文本检索或语义相似度等文本embedding任务,
重要的是分数的相对顺序而非绝对值,
因此这不应成为问题。
引用
如果您觉得我们的论文或模型有帮助,请考虑如下引用:
@article{wang2022text,
title={Text Embeddings by Weakly-Supervised Contrastive Pre-training},
author={Wang, Liang and Yang, Nan and Huang, Xiaolong and Jiao, Binxing and Yang, Linjun and Jiang, Daxin and Majumder, Rangan and Wei, Furu},
journal={arXiv preprint arXiv:2212.03533},
year={2022}
}
局限性
该模型仅适用于英文文本。长文本将被截断至最多512个token。
正在翻译中,请稍候...