e5 large v2

提供商intfloat
分类sentence-similarity
许可证mit
下载量1.1K
星标0

简介

e5-large-v2 是一款高性能的文本向量化模型，专注于将自然语言转化为高质量的语义向量。对于开发者而言，它不是用来聊天的对话模型，而是构建 RAG（检索增强生成）系统的核心组件。它在语义相似度计算和文本检索方面表现出色，能有效将用户查询与知识库中的相关片段精准匹配。由于其基于 MIT 协议开源且推理开销相对可控，非常适合部署在本地环境，作为向量数据库的前端编码器，替代或优化传统的关键词检索。

核心亮点

  • 语义检索核心，大幅提升 RAG 知识库召回率
  • MIT 协议开源，支持私有化部署且无商用压力
  • 专注文本向量化，完美适配各类向量数据库
  • 语义匹配精准，有效解决关键词不匹配的痛点

使用方法

安装依赖
# 安装 Hugging Face transformers
pip install transformers torch
SDK 使用
# 使用 transformers 加载模型
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("intfloat/e5-large-v2")
tokenizer = AutoTokenizer.from_pretrained("intfloat/e5-large-v2")

Hugging Face 下载

我们推荐使用命令行或者 Hugging Face Hub SDK 来进行模型的下载。

操作指引：在下载前，请先通过如下命令安装 huggingface_hub：

操作指引 
pip install -U huggingface_hub

命令行下载

下载完整模型库

下载完整模型库 
huggingface-cli download intfloat/e5-large-v2

下载单个文件到指定本地文件夹（以下载 config.json 到当前路径下 ./dir 目录为例）

下载单个文件到指定本地文件夹（以下载 config.json 到当前路径下 ./dir 目录为例） 
huggingface-cli download intfloat/e5-large-v2 config.json --local-dir ./dir

更多命令行下载选项，可参见官方文档

SDK 下载

SDK 下载 
# 模型下载
from huggingface_hub import snapshot_download
model_dir = snapshot_download('intfloat/e5-large-v2')

Git 下载

请确保 lfs 已经被正确安装

Git 下载 
git lfs install
git clone https://huggingface.co/intfloat/e5-large-v2

如果您希望跳过 lfs 大文件下载，可以使用如下命令

跳过 LFS 
GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/intfloat/e5-large-v2

模型文件托管在 Hugging Face Hub，使用 HF CLI / SDK / Git 直接下载，不经过本站。

PyTorch / Transformers 使用

安装 Transformers

安装 Transformers 
pip install -U transformers torch

模型加载和推理

模型加载和推理 
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained('intfloat/e5-large-v2')
tokenizer = AutoTokenizer.from_pretrained('intfloat/e5-large-v2')

模型下载

我们推荐使用命令行或者 ModelScope SDK 来进行模型的下载。

操作指引：在下载前，请先通过如下命令安装 ModelScope：

操作指引 
pip install modelscope

命令行下载

下载完整模型库

下载完整模型库 
modelscope download --model intfloat/e5-large-v2

下载单个文件到指定本地文件夹（以下载 README.md 到当前路径下 dir 目录为例）

下载单个文件到指定本地文件夹（以下载 README.md 到当前路径下 dir 目录为例） 
modelscope download --model intfloat/e5-large-v2 README.md --local_dir ./dir

更多更丰富的命令行下载选项，可参见具体文档

SDK 下载

SDK 下载 
# 模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('intfloat/e5-large-v2')

Git 下载

请确保 lfs 已经被正确安装

Git 下载 
git lfs install
git clone https://www.modelscope.cn/intfloat/e5-large-v2.git

如果您希望跳过 lfs 大文件下载，可以使用如下命令

跳过 LFS 
GIT_LFS_SKIP_SMUDGE=1 git clone https://www.modelscope.cn/intfloat/e5-large-v2.git

ModelScope 模型页直接下载模型文件；无需将模型文件放在本站服务器。

Notebook 快速开发

下载并安装 ModelScope library

下载并安装 ModelScope library 
pip install "modelscope[audio,cv,nlp,multi-modal,science]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html

模型加载和推理

模型加载和推理 
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks

p = pipeline('text-generation', 'intfloat/e5-large-v2')

完整文档

来源: HuggingFace

---
tags:

  • mteb

  • Sentence Transformers

  • sentence-similarity

  • sentence-transformers

model-index:
  • name: e5-large-v2

results:
- task:
type: Classification
dataset:
type: mteb/amazon_counterfactual
name: MTEB AmazonCounterfactualClassification (en)
config: en
split: test
revision: e8379541af4e31359cca9fbcf4b00f2671dba205
metrics:
- type: accuracy
value: 79.22388059701493
- type: ap
value: 43.20816505595132
- type: f1
value: 73.27811303522058
- task:
type: Classification
dataset:
type: mteb/amazon_polarity
name: MTEB AmazonPolarityClassification
config: default
split: test
revision: e2d317d38cd51312af73b3d32a06d1a08b442046
metrics:
- type: accuracy
value: 93.748325
- type: ap
value: 90.72534979701297
- type: f1
value: 93.73895874282185
- task:
type: Classification
dataset:
type: mteb/amazon_reviews_multi
name: MTEB AmazonReviewsClassification (en)
config: en
split: test
revision: 1399c76144fd37290681b995c656ef9b2e06e26d
metrics:
- type: accuracy
value: 48.612
- type: f1
value: 47.61157345898393
- task:
type: Retrieval
dataset:
type: arguana
name: MTEB ArguAna
config: default
split: test
revision: None
metrics:
- type: map_at_1
value: 23.541999999999998
- type: map_at_10
value: 38.208
- type: map_at_100
value: 39.417
- type: map_at_1000
value: 39.428999999999995
- type: map_at_3
value: 33.95
- type: map_at_5
value: 36.329
- type: mrr_at_1
value: 23.755000000000003
- type: mrr_at_10
value: 38.288
- type: mrr_at_100
value: 39.511
- type: mrr_at_1000
value: 39.523
- type: mrr_at_3
value: 34.009
- type: mrr_at_5
value: 36.434
- type: ndcg_at_1
value: 23.541999999999998
- type: ndcg_at_10
value: 46.417
- type: ndcg_at_100
value: 51.812000000000005
- type: ndcg_at_1000
value: 52.137
- type: ndcg_at_3
value: 37.528
- type: ndcg_at_5
value: 41.81
- type: precision_at_1
value: 23.541999999999998
- type: precision_at_10
value: 7.269
- type: precision_at_100
value: 0.9690000000000001
- type: precision_at_1000
value: 0.099
- type: precision_at_3
value: 15.979
- type: precision_at_5
value: 11.664
- type: recall_at_1
value: 23.541999999999998
- type: recall_at_10
value: 72.688
- type: recall_at_100
value: 96.871
- type: recall_at_1000
value: 99.431
- type: recall_at_3
value: 47.937000000000005
- type: recall_at_5
value: 58.321
- task:
type: Clustering
dataset:
type: mteb/arxiv-clustering-p2p
name: MTEB ArxivClusteringP2P
config: default
split: test
revision: a122ad7f3f0291bf49cc6f4d32aa80929df69d5d
metrics:
- type: v_measure
value: 45.546499570522094
- task:
type: Clustering
dataset:
type: mteb/arxiv-clustering-s2s
name: MTEB ArxivClusteringS2S
config: default
split: test
revision: f910caf1a6075f7329cdf8c1a6135696f37dbd53
metrics:
- type: v_measure
value: 41.01607489943561
- task:
type: Reranking
dataset:
type: mteb/askubuntudupquestions-reranking
name: MTEB AskUbuntuDupQuestions
config: default
split: test
revision: 2000358ca161889fa9c082cb41daa8dcfb161a54
metrics:
- type: map
value: 59.616107510107774
- type: mrr
value: 72.75106626214661
- task:
type: STS
dataset:
type: mteb/biosses-sts
name: MTEB BIOSSES
config: default
split: test
revision: d3fb88f8f02e40887cd149695127462bbcf29b4a
metrics:
- type: cos_sim_pearson
value: 84.33018094733868
- type: cos_sim_spearman
value: 83.60190492611737
- type: euclidean_pearson
value: 82.1492450218961
- type: euclidean_spearman
value: 82.70308926526991
- type: manhattan_pearson
value: 81.93959600076842
- type: manhattan_spearman
value: 82.73260801016369
- task:
type: Classification
dataset:
type: mteb/banking77
name: MTEB Banking77Classification
config: default
split: test
revision: 0fd18e25b25c072e09e0d92ab615fda904d66300
metrics:
- type: accuracy
value: 84.54545454545455
- type: f1
value: 84.49582530928923
- task:
type: Clustering
dataset:
type: mteb/biorxiv-clustering-p2p
name: MTEB BiorxivClusteringP2P
config: default
split: test
revision: 65b79d1d13f80053f67aca9498d9402c2d9f1f40
metrics:
- type: v_measure
value: 37.362725540120096
- task:
type: Clustering
dataset:
type: mteb/biorxiv-clustering-s2s
name: MTEB BiorxivClusteringS2S
config: default
split: test
revision: 258694dd0231531bc1fd9de6ceb52a0853c6d908
metrics:
- type: v_measure
value: 34.849509608178145
- task:
type: Retrieval
dataset:
type: BeIR/cqadupstack
name: MTEB CQADupstackAndroidRetrieval
config: default
split: test
revision: None
metrics:
- type: map_at_1
value: 31.502999999999997
- type: map_at_10
value: 43.323
- type: map_at_100
value: 44.708999999999996
- type: map_at_1000
value: 44.838
- type: map_at_3
value: 38.987
- type: map_at_5
value: 41.516999999999996
- type: mrr_at_1
value: 38.769999999999996
- type: mrr_at_10
value: 49.13
- type: mrr_at_100
value: 49.697
- type: mrr_at_1000
value: 49.741
- type: mrr_at_3
value: 45.804
- type: mrr_at_5
value: 47.842
- type: ndcg_at_1
value: 38.769999999999996
- type: ndcg_at_10
value: 50.266999999999996
- type: ndcg_at_100
value: 54.967
- type: ndcg_at_1000
value: 56.976000000000006
- type: ndcg_at_3
value: 43.823
- type: ndcg_at_5
value: 47.12
- type: precision_at_1
value: 38.769999999999996
- type: precision_at_10
value: 10.057
- type: precision_at_100
value: 1.554
- type: precision_at_1000
value: 0.202
- type: precision_at_3
value: 21.125
- type: precision_at_5
value: 15.851
- type: recall_at_1
value: 31.502999999999997
- type: recall_at_10
value: 63.715999999999994
- type: recall_at_100
value: 83.61800000000001
- type: recall_at_1000
value: 96.63199999999999
- type: recall_at_3
value: 45.403
- type: recall_at_5
value: 54.481
- task:
type: Retrieval
dataset:
type: BeIR/cqadupstack
name: MTEB CQADupstackEnglishRetrieval
config: default
split: test
revision: None
metrics:
- type: map_at_1
value: 27.833000000000002
- type: map_at_10
value: 37.330999999999996
- type: map_at_100
value: 38.580999999999996
- type: map_at_1000
value: 38.708
- type: map_at_3
value: 34.713
- type: map_at_5
value: 36.104
- type: mrr_at_1
value: 35.223
- type: mrr_at_10
value: 43.419000000000004
- type: mrr_at_100
value: 44.198
- type: mrr_at_1000
value: 44.249
- type: mrr_at_3
value: 41.614000000000004
- type: mrr_at_5
value: 42.553000000000004
- type: ndcg_at_1
value: 35.223
- type: ndcg_at_10
value: 42.687999999999995
- type: ndcg_at_100
value: 47.447
- type: ndcg_at_1000
value: 49.701
- type: ndcg_at_3
value: 39.162
- type: ndcg_at_5
value: 40.557
- type: precision_at_1
value: 35.223
- type: precision_at_10
value: 7.962
- type: precision_a

查看官方来源