Langcache Embed V2

🚀 基于 redis/langcache-embed-v1 的句子转换器

该项目基于 sentence-transformers 框架，在三元组数据集上对 redis/langcache-embed-v1 模型进行微调。它能够将句子和段落映射到 768 维的密集向量空间，可用于语义文本相似度计算、语义搜索、释义挖掘、文本分类、聚类等任务。

🚀 快速开始

安装 Sentence Transformers 库

首先，你需要安装 Sentence Transformers 库：

pip install -U sentence-transformers

加载模型并进行推理

安装完成后，你可以加载此模型并进行推理：

from sentence_transformers import SentenceTransformer

# 从 Hugging Face Hub 下载模型
model = SentenceTransformer("redis/langcache-embed-v2")
# 进行推理
sentences = [
    'What are some examples of crimes understood as a moral turpitude?',
    'What are some examples of crimes of moral turpitude?',
    'What are some examples of crimes understood as a legal aptitude?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# 获取嵌入向量的相似度分数
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

✨ 主要特性

• 高维向量映射：能够将句子和段落映射到 768 维的密集向量空间，便于进行语义分析。
• 多任务支持：可用于语义文本相似度计算、语义搜索、释义挖掘、文本分类、聚类等多种自然语言处理任务。

📦 安装指南

安装 Sentence Transformers 库：

pip install -U sentence-transformers

💻 使用示例

基础用法

from sentence_transformers import SentenceTransformer

# 从 Hugging Face Hub 下载模型
model = SentenceTransformer("redis/langcache-embed-v2")
# 进行推理
sentences = [
    'What are some examples of crimes understood as a moral turpitude?',
    'What are some examples of crimes of moral turpitude?',
    'What are some examples of crimes understood as a legal aptitude?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# 获取嵌入向量的相似度分数
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

📚 详细文档

模型详情

模型描述

模型来源

• 文档：Sentence Transformers 文档
• 仓库：GitHub 上的 Sentence Transformers
• Hugging Face：Hugging Face 上的 Sentence Transformers

完整模型架构

SentenceTransformer(
  (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: ModernBertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)

训练详情

• 数据集：三元组数据集
• 大小：36,864 个训练样本
• 列：anchor、positive、negative_1、negative_2 和 negative_3

样本示例

损失函数

使用 MatryoshkaLoss，参数如下：

{
    "loss": "CachedMultipleNegativesRankingLoss",
    "matryoshka_dims": [768,512,256,128,64],
    "matryoshka_weights": [1,1,1,1,1],
    "n_dims_per_step": -1
}

评估

🔧 技术细节

训练日志

框架版本

• Python：3.11.11
• Sentence Transformers：4.1.0
• Transformers：4.51.3
• PyTorch：2.6.0+cu124
• Accelerate：1.6.0
• Datasets：3.5.1
• Tokenizers：0.21.1

📄 许可证

文档中未提及相关许可证信息。

📚 引用

Redis Langcache-embed 模型

如果您使用了我们的模型或基于我们的研究成果进行开发，我们鼓励您引用我们的工作：

@inproceedings{langcache-embed-v1,
    title = "Advancing Semantic Caching for LLMs with Domain-Specific Embeddings and Synthetic Data",
    author = "Gill, Cechmanek, Hutcherson, Rajamohan, Agarwal, Gulzar, Singh, Dion",
    month = "04",
    year = "2025",
    url = "https://arxiv.org/abs/2504.02268",
}

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

@misc{gao2021scaling,
    title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
    author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
    year={2021},
    eprint={2101.06983},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}