标签:

• 句子转换器
• 交叉编码器
• 训练生成
• 数据集大小:22258
• 损失函数:FitMixinLoss 基础模型: cross-encoder/ms-marco-MiniLM-L6-v2 管道标签: 文本排序 库名称: sentence-transformers 评估指标:
• 平均精度(map)
• 前10命中率(mrr@10)
• 标准化折损累积增益(ndcg@10) 模型索引:
• 名称: 基于cross-encoder/ms-marco-MiniLM-L6-v2的交叉编码器 结果:
  • 任务: 类型: 交叉编码器重排序 名称: 交叉编码器重排序 数据集: 名称: 交叉重排序开发集混合负样本 类型: cross-rerank-dev-mixed-neg 指标:
    • 类型: 平均精度(map) 值: 0.4873053613053613 名称: 平均精度
    • 类型: 前10命中率(mrr@10) 值: 0.48394871794871797 名称: 前10命中率
    • 类型: 标准化折损累积增益(ndcg@10) 值: 0.5970778430138177 名称: 标准化折损累积增益

基于cross-encoder/ms-marco-MiniLM-L6-v2的交叉编码器

这是一个基于cross-encoder/ms-marco-MiniLM-L6-v2微调的交叉编码器模型，使用sentence-transformers库训练。它计算文本对的分数，可用于文本重排序和语义搜索。

模型详情

模型描述

• 模型类型: 交叉编码器
• 基础模型: cross-encoder/ms-marco-MiniLM-L6-v2
• 最大序列长度: 512个标记
• 输出标签数量: 1个标签

模型来源

• 文档: Sentence Transformers文档
• 文档: 交叉编码器文档
• 代码库: GitHub上的Sentence Transformers
• Hugging Face: Hugging Face上的交叉编码器

使用方法

直接使用(Sentence Transformers)

首先安装Sentence Transformers库:

pip install -U sentence-transformers

然后可以加载此模型并进行推理。

from sentence_transformers import CrossEncoder

# 从Hugging Face Hub下载
model = CrossEncoder("CharlesPing/finetuned-cross-encoder-l6-v2")
# 获取文本对的分数
pairs = [
    ['‘Getting hung up on the exact nature of the records is interesting, and there’s lots of technical work that can be done there, but the main take-home response there is that the trends we’ve been seeing since the 1970s are continuing and have not paused in any way,’ he said.”', 'Rosenzweig also criticized the "waffling—encouraged by the NPOV policy—[which] means that it is hard to discern any overall interpretive stance in Wikipedia history".'],
    ['After the 9/11 terrorist attacks grounded commercial air traffic, "there was a temperature drop while the airplanes weren\'t flying, for the week afterwards."', 'Play media At 9:42\xa0a.m., the Federal Aviation Administration (FAA) grounded all civilian aircraft within the continental U.S., and civilian aircraft already in flight were told to land immediately.'],
    ['But the central message of the IPCC AR4, is confirmed by the peer reviewed literature.', 'Scientific consensus is normally achieved through communication at conferences, publication in the scientific literature, replication (reproducible results by others), and peer review.'],
    ['"Many people think the science of climate change is settled.', 'During his administration, the bridge from Filadelfia and Liberia was constructed, as was the Old National Theater.'],
    ['“Even if you could calculate some sort of meaningful global temperature statistic, the figure would be unimportant.', 'Quantitative information or data is based on quantities obtained using a quantifiable measurement process.'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)

# 或根据与单个文本的相似性对不同文本进行排序
ranks = model.rank(
    '‘Getting hung up on the exact nature of the records is interesting, and there’s lots of technical work that can be done there, but the main take-home response there is that the trends we’ve been seeing since the 1970s are continuing and have not paused in any way,’ he said.”',
    [
        'Rosenzweig also criticized the "waffling—encouraged by the NPOV policy—[which] means that it is hard to discern any overall interpretive stance in Wikipedia history".',
        'Play media At 9:42\xa0a.m., the Federal Aviation Administration (FAA) grounded all civilian aircraft within the continental U.S., and civilian aircraft already in flight were told to land immediately.',
        'Scientific consensus is normally achieved through communication at conferences, publication in the scientific literature, replication (reproducible results by others), and peer review.',
        'During his administration, the bridge from Filadelfia and Liberia was constructed, as was the Old National Theater.',
        'Quantitative information or data is based on quantities obtained using a quantifiable measurement process.',
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

评估

指标

交叉编码器重排序

• 数据集: cross-rerank-dev-mixed-neg
• 使用CrossEncoderRerankingEvaluator评估，参数如下:

  {
      "at_k": 10
  }
  

训练详情

训练数据集

未命名数据集

• 大小: 22,258个训练样本
• 列: sentence_0, sentence_1和label
• 基于前1000个样本的近似统计:

  	sentence_0	sentence_1	label
  类型	字符串	字符串	浮点数
  详情	最小: 26字符 平均: 121.91字符 最大: 319字符	最小: 36字符 平均: 140.85字符 最大: 573字符	最小: 0.0 平均: 0.16 最大: 1.0

• 样本:

  sentence_0	sentence_1	label
  ‘Getting hung up on the exact nature of the records is interesting, and there’s lots of technical work that can be done there, but the main take-home response there is that the trends we’ve been seeing since the 1970s are continuing and have not paused in any way,’ he said.”	Rosenzweig also criticized the "waffling—encouraged by the NPOV policy—[which] means that it is hard to discern any overall interpretive stance in Wikipedia history".	1.0
  After the 9/11 terrorist attacks grounded commercial air traffic, "there was a temperature drop while the airplanes weren't flying, for the week afterwards."	Play media At 9:42 a.m., the Federal Aviation Administration (FAA) grounded all civilian aircraft within the continental U.S., and civilian aircraft already in flight were told to land immediately.	1.0
  But the central message of the IPCC AR4, is confirmed by the peer reviewed literature.	Scientific consensus is normally achieved through communication at conferences, publication in the scientific literature, replication (reproducible results by others), and peer review.	1.0

• 损失函数: FitMixinLoss

训练超参数

非默认超参数

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16

训练日志

框架版本

• Python: 3.11.12
• 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

引用

BibTeX

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",
}