Reranker Ms Marco MiniLM L6 V2 Gooaq Bce
这是一个从cross-encoder/ms-marco-MiniLM-L6-v2微调而来的交叉编码器模型,使用sentence-transformers库开发。它能够计算文本对的得分,可用于文本重排序和语义搜索。
下载量 15
发布时间 : 3/30/2025
模型简介
该模型是一个基于MiniLM架构的交叉编码器,专门用于文本重排序任务。它通过计算查询和文档之间的相关性得分,优化搜索结果的排序质量。
模型特点
高效重排序
专门优化用于重排序任务,能够显著提升搜索结果的相关性
多数据集验证
在GooAQ、MSMARCO、NFCorpus和NQ等多个数据集上进行了验证,表现稳定
长文本处理
支持最大512个标记的序列长度,适合处理较长的查询和文档
模型能力
文本相关性评分
搜索结果重排序
语义搜索优化
使用案例
信息检索
搜索引擎结果优化
对初步检索结果进行重排序,提高相关文档的排名
在GooAQ开发集上达到0.6822的NDCG@10分数
问答系统
对候选答案进行相关性排序,选择最匹配的答案
在NanoNQ数据集上达到0.5091的NDCG@10分数
医疗健康
医疗问答匹配
匹配用户医疗问题与专业医学解答
如示例中所示,能准确识别与左臂疼痛相关的医学解释
🚀 基于GooAQ训练的ModernBERT-base模型
这是一个基于Cross Encoder的模型,它使用sentence-transformers库,从cross-encoder/ms-marco-MiniLM-L6-v2微调而来。该模型可以计算文本对的得分,可用于文本重排序和语义搜索。
🚀 快速开始
此模型可用于计算文本对的得分,适用于文本重排序和语义搜索任务。下面将介绍如何使用此模型。
✨ 主要特性
- 文本重排序:能够为文本对计算得分,从而实现文本重排序。
- 语义搜索:可用于语义搜索任务,帮助用户更精准地找到相关文本。
📦 安装指南
首先,你需要安装Sentence Transformers库:
pip install -U sentence-transformers
💻 使用示例
基础用法
安装好库之后,你可以加载此模型并进行推理:
from sentence_transformers import CrossEncoder
# 从🤗 Hub下载模型
model = CrossEncoder("ayushexel/reranker-ms-marco-MiniLM-L6-v2-gooaq-bce")
# 获取文本对的得分
pairs = [
['what does it mean when you get a sharp pain in your left arm?', 'Pain in the left arm A pain in your left arm could mean you have a bone or joint injury, a pinched nerve, or a problem with your heart. Read on to learn more about the causes of left arm pain and what symptoms could signal a serious problem.'],
['what does it mean when you get a sharp pain in your left arm?', "In this Article Whether it's throbbing, aching, or sharp, everyone has been in pain. The uncomfortable sensation is a red flag. Pain in your armpit could mean that you've simply strained a muscle, which is eased with ice and rest. It could also be a sign of more serious conditions, like an infection or breast cancer."],
['what does it mean when you get a sharp pain in your left arm?', 'Sharp: When you feel a sudden, intense spike of pain, that qualifies as “sharp.” Sharp pain may also fit the descriptors cutting and shooting. Stabbing: Like sharp pain, stabbing pain occurs suddenly and intensely. However, stabbing pain may fade and reoccur many times.'],
['what does it mean when you get a sharp pain in your left arm?', 'Symptoms. A herniated disc in the neck can cause neck pain, radiating arm pain, shoulder pain, and numbness or tingling in the arm or hand. The quality and type of pain can vary from dull, aching, and difficult to localize to sharp, burning, and easy to pinpoint.'],
['what does it mean when you get a sharp pain in your left arm?', 'Injuries or trauma to any part of the arm or shoulder, including bone fractures, joint dislocations, and muscle strains and sprains, are common causes of arm pain. Sometimes diseases that affect other organs in the body, like peripheral vascular disease or arthritis, can be the cause of pain in the arm.'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)
# 或者根据与单个文本的相似度对不同文本进行排序
ranks = model.rank(
'what does it mean when you get a sharp pain in your left arm?',
[
'Pain in the left arm A pain in your left arm could mean you have a bone or joint injury, a pinched nerve, or a problem with your heart. Read on to learn more about the causes of left arm pain and what symptoms could signal a serious problem.',
"In this Article Whether it's throbbing, aching, or sharp, everyone has been in pain. The uncomfortable sensation is a red flag. Pain in your armpit could mean that you've simply strained a muscle, which is eased with ice and rest. It could also be a sign of more serious conditions, like an infection or breast cancer.",
'Sharp: When you feel a sudden, intense spike of pain, that qualifies as “sharp.” Sharp pain may also fit the descriptors cutting and shooting. Stabbing: Like sharp pain, stabbing pain occurs suddenly and intensely. However, stabbing pain may fade and reoccur many times.',
'Symptoms. A herniated disc in the neck can cause neck pain, radiating arm pain, shoulder pain, and numbness or tingling in the arm or hand. The quality and type of pain can vary from dull, aching, and difficult to localize to sharp, burning, and easy to pinpoint.',
'Injuries or trauma to any part of the arm or shoulder, including bone fractures, joint dislocations, and muscle strains and sprains, are common causes of arm pain. Sometimes diseases that affect other organs in the body, like peripheral vascular disease or arthritis, can be the cause of pain in the arm.',
]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]
📚 详细文档
模型详情
模型描述
属性 | 详情 |
---|---|
模型类型 | 交叉编码器(Cross Encoder) |
基础模型 | cross-encoder/ms-marco-MiniLM-L6-v2 |
最大序列长度 | 512个词元 |
输出标签数量 | 1个标签 |
语言 | 英语 |
许可证 | apache-2.0 |
模型来源
- 文档:Sentence Transformers文档
- 文档:Cross Encoder文档
- 代码仓库:GitHub上的Sentence Transformers
- Hugging Face:Hugging Face上的Cross Encoders
评估
指标
交叉编码器重排序(数据集:gooaq-dev
)
使用CrossEncoderRerankingEvaluator
进行评估,参数如下:
{
"at_k": 10,
"always_rerank_positives": false
}
指标 | 值 |
---|---|
map | 0.6380 (+0.2121) |
mrr@10 | 0.6361 (+0.2199) |
ndcg@10 | 0.6822 (+0.2001) |
交叉编码器重排序(数据集:NanoMSMARCO_R100
、NanoNFCorpus_R100
和NanoNQ_R100
)
使用CrossEncoderRerankingEvaluator
进行评估,参数如下:
{
"at_k": 10,
"always_rerank_positives": true
}
指标 | NanoMSMARCO_R100 | NanoNFCorpus_R100 | NanoNQ_R100 |
---|---|---|---|
map | 0.5437 (+0.0541) | 0.3885 (+0.1275) | 0.4626 (+0.0430) |
mrr@10 | 0.5348 (+0.0573) | 0.5630 (+0.0632) | 0.4628 (+0.0361) |
ndcg@10 | 0.6060 (+0.0655) | 0.4077 (+0.0827) | 0.5091 (+0.0084) |
交叉编码器Nano BEIR(数据集:NanoBEIR_R100_mean
)
使用CrossEncoderNanoBEIREvaluator
进行评估,参数如下:
{
"dataset_names": [
"msmarco",
"nfcorpus",
"nq"
],
"rerank_k": 100,
"at_k": 10,
"always_rerank_positives": true
}
指标 | 值 |
---|---|
map | 0.4649 (+0.0749) |
mrr@10 | 0.5202 (+0.0522) |
ndcg@10 | 0.5076 (+0.0522) |
训练详情
训练数据集
未命名数据集
- 规模:2,223,773个训练样本
- 列名:
question
、answer
和label
- 基于前1000个样本的近似统计信息:
| | 问题 | 答案 | 标签 |
|------|------|------|------|
| 类型 | 字符串 | 字符串 | 整数 |
| 详情 |
- 最小:19个字符
- 平均:45.87个字符
- 最大:88个字符
- 最小:61个字符
- 平均:253.13个字符
- 最大:374个字符
- 0:约86.70%
- 1:约13.30%
- 样本示例:
| 问题 | 答案 | 标签 |
|------|------|------|
|
what does it mean when you get a sharp pain in your left arm?
|Pain in the left arm A pain in your left arm could mean you have a bone or joint injury, a pinched nerve, or a problem with your heart. Read on to learn more about the causes of left arm pain and what symptoms could signal a serious problem.
|1
| |what does it mean when you get a sharp pain in your left arm?
|In this Article Whether it's throbbing, aching, or sharp, everyone has been in pain. The uncomfortable sensation is a red flag. Pain in your armpit could mean that you've simply strained a muscle, which is eased with ice and rest. It could also be a sign of more serious conditions, like an infection or breast cancer.
|0
| |what does it mean when you get a sharp pain in your left arm?
|Sharp: When you feel a sudden, intense spike of pain, that qualifies as “sharp.” Sharp pain may also fit the descriptors cutting and shooting. Stabbing: Like sharp pain, stabbing pain occurs suddenly and intensely. However, stabbing pain may fade and reoccur many times.
|0
| - 损失函数:
BinaryCrossEntropyLoss
,参数如下:
{
"activation_fn": "torch.nn.modules.linear.Identity",
"pos_weight": 7
}
训练超参数
非默认超参数
eval_strategy
:按步骤评估per_device_train_batch_size
:2048per_device_eval_batch_size
:2048learning_rate
:2e-05warmup_ratio
:0.1seed
:12bf16
:Truedataloader_num_workers
:12load_best_model_at_end
:True
所有超参数
点击展开
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: stepsprediction_loss_only
: Trueper_device_train_batch_size
: 2048per_device_eval_batch_size
: 2048per_gpu_train_batch_size
: Noneper_gpu_eval_batch_size
: Nonegradient_accumulation_steps
: 1eval_accumulation_steps
: Nonetorch_empty_cache_steps
: Nonelearning_rate
: 2e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1.0num_train_epochs
: 3max_steps
: -1lr_scheduler_type
: linearlr_scheduler_kwargs
: {}warmup_ratio
: 0.1warmup_steps
: 0log_level
: passivelog_level_replica
: warninglog_on_each_node
: Truelogging_nan_inf_filter
: Truesave_safetensors
: Truesave_on_each_node
: Falsesave_only_model
: Falserestore_callback_states_from_checkpoint
: Falseno_cuda
: Falseuse_cpu
: Falseuse_mps_device
: Falseseed
: 12data_seed
: Nonejit_mode_eval
: Falseuse_ipex
: Falsebf16
: Truefp16
: Falsefp16_opt_level
: O1half_precision_backend
: autobf16_full_eval
: Falsefp16_full_eval
: Falsetf32
: Nonelocal_rank
: 0ddp_backend
: Nonetpu_num_cores
: Nonetpu_metrics_debug
: Falsedebug
: []dataloader_drop_last
: Falsedataloader_num_workers
: 12dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Trueignore_data_skip
: Falsefsdp
: []fsdp_min_num_params
: 0fsdp_config
: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}tp_size
: 0fsdp_transformer_layer_cls_to_wrap
: Noneaccelerator_config
: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}deepspeed
: Nonelabel_smoothing_factor
: 0.0optim
: adamw_torchoptim_args
: Noneadafactor
: Falsegroup_by_length
: Falselength_column_name
: lengthddp_find_unused_parameters
: Noneddp_bucket_cap_mb
: Noneddp_broadcast_buffers
: Falsedataloader_pin_memory
: Truedataloader_persistent_workers
: Falseskip_memory_metrics
: Trueuse_legacy_prediction_loop
: Falsepush_to_hub
: Falseresume_from_checkpoint
: Nonehub_model_id
: Nonehub_strategy
: every_savehub_private_repo
: Nonehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseinclude_for_metrics
: []eval_do_concat_batches
: Truefp16_backend
: autopush_to_hub_model_id
: Nonepush_to_hub_organization
: Nonemp_parameters
:auto_find_batch_size
: Falsefull_determinism
: Falsetorchdynamo
: Noneray_scope
: lastddp_timeout
: 1800torch_compile
: Falsetorch_compile_backend
: Nonetorch_compile_mode
: Nonedispatch_batches
: Nonesplit_batches
: Noneinclude_tokens_per_second
: Falseinclude_num_input_tokens_seen
: Falseneftune_noise_alpha
: Noneoptim_target_modules
: Nonebatch_eval_metrics
: Falseeval_on_start
: Falseuse_liger_kernel
: Falseeval_use_gather_object
: Falseaverage_tokens_across_devices
: Falseprompts
: Nonebatch_sampler
: batch_samplermulti_dataset_batch_sampler
: proportional
训练日志
轮数 | 步骤 | 训练损失 | gooaq-dev_ndcg@10 | NanoMSMARCO_R100_ndcg@10 | NanoNFCorpus_R100_ndcg@10 | NanoNQ_R100_ndcg@10 | NanoBEIR_R100_mean_ndcg@10 |
---|---|---|---|---|---|---|---|
-1 | -1 | - | 0.6371 (+0.1550) | 0.6686 (+0.1282) | 0.3930 (+0.0680) | 0.7599 (+0.2592) | 0.6072 (+0.1518) |
0.0009 | 1 | 2.1175 | - | - | - | - | - |
0.1842 | 200 | 1.1892 | - | - | - | - | - |
0.3683 | 400 | 0.676 | - | - | - | - | - |
0.5525 | 600 | 0.6268 | - | - | - | - | - |
0.7366 | 800 | 0.606 | - | - | - | - | - |
0.9208 | 1000 | 0.5933 | 0.6731 (+0.1910) | 0.6038 (+0.0634) | 0.4572 (+0.1321) | 0.5220 (+0.0213) | 0.5277 (+0.0723) |
1.1050 | 1200 | 0.5756 | - | - | - | - | - |
1.2891 | 1400 | 0.5625 | - | - | - | - | - |
1.4733 | 1600 | 0.5575 | - | - | - | - | - |
1.6575 | 1800 | 0.549 | - | - | - | - | - |
1.8416 | 2000 | 0.5475 | 0.6799 (+0.1977) | 0.6072 (+0.0667) | 0.4278 (+0.1028) | 0.5031 (+0.0024) | 0.5127 (+0.0573) |
2.0258 | 2200 | 0.5391 | - | - | - | - | - |
2.2099 | 2400 | 0.5276 | - | - | - | - | - |
2.3941 | 2600 | 0.5271 | - | - | - | - | - |
2.5783 | 2800 | 0.5264 | - | - | - | - | - |
2.7624 | 3000 | 0.5244 | 0.6822 (+0.2001) | 0.6060 (+0.0655) | 0.4077 (+0.0827) | 0.5091 (+0.0084) | 0.5076 (+0.0522) |
2.9466 | 3200 | 0.5235 | - | - | - | - | - |
-1 | -1 | - | 0.6822 (+0.2001) | 0.6060 (+0.0655) | 0.4077 (+0.0827) | 0.5091 (+0.0084) | 0.5076 (+0.0522) |
注:加粗行表示保存的检查点。
框架版本
- Python:3.11.0
- Sentence Transformers:4.0.1
- Transformers:4.50.3
- PyTorch:2.6.0+cu124
- Accelerate:1.5.2
- Datasets:3.5.0
- Tokenizers:0.21.1
📄 许可证
本模型使用apache-2.0许可证。
🔖 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",
}
Jina Embeddings V3
Jina Embeddings V3 是一个多语言句子嵌入模型,支持超过100种语言,专注于句子相似度和特征提取任务。
文本嵌入
Transformers 支持多种语言

J
jinaai
3.7M
911
Ms Marco MiniLM L6 V2
Apache-2.0
基于MS Marco段落排序任务训练的交叉编码器模型,用于信息检索中的查询-段落相关性评分
文本嵌入 英语
M
cross-encoder
2.5M
86
Opensearch Neural Sparse Encoding Doc V2 Distill
Apache-2.0
基于蒸馏技术的稀疏检索模型,专为OpenSearch优化,支持免推理文档编码,在搜索相关性和效率上优于V1版本
文本嵌入
Transformers 英语

O
opensearch-project
1.8M
7
Sapbert From PubMedBERT Fulltext
Apache-2.0
基于PubMedBERT的生物医学实体表征模型,通过自对齐预训练优化语义关系捕捉
文本嵌入 英语
S
cambridgeltl
1.7M
49
Gte Large
MIT
GTE-Large 是一个强大的句子转换器模型,专注于句子相似度和文本嵌入任务,在多个基准测试中表现出色。
文本嵌入 英语
G
thenlper
1.5M
278
Gte Base En V1.5
Apache-2.0
GTE-base-en-v1.5 是一个英文句子转换器模型,专注于句子相似度任务,在多个文本嵌入基准测试中表现优异。
文本嵌入
Transformers 支持多种语言

G
Alibaba-NLP
1.5M
63
Gte Multilingual Base
Apache-2.0
GTE Multilingual Base 是一个多语言的句子嵌入模型,支持超过50种语言,适用于句子相似度计算等任务。
文本嵌入
Transformers 支持多种语言

G
Alibaba-NLP
1.2M
246
Polybert
polyBERT是一个化学语言模型,旨在实现完全由机器驱动的超快聚合物信息学。它将PSMILES字符串映射为600维密集指纹,以数值形式表示聚合物化学结构。
文本嵌入
Transformers

P
kuelumbus
1.0M
5
Bert Base Turkish Cased Mean Nli Stsb Tr
Apache-2.0
基于土耳其语BERT的句子嵌入模型,专为语义相似度任务优化
文本嵌入
Transformers 其他

B
emrecan
1.0M
40
GIST Small Embedding V0
MIT
基于BAAI/bge-small-en-v1.5模型微调的文本嵌入模型,通过MEDI数据集与MTEB分类任务数据集训练,优化了检索任务的查询编码能力。
文本嵌入
Safetensors 英语
G
avsolatorio
945.68k
29
精选推荐AI模型
Llama 3 Typhoon V1.5x 8b Instruct
专为泰语设计的80亿参数指令模型,性能媲美GPT-3.5-turbo,优化了应用场景、检索增强生成、受限生成和推理任务
大型语言模型
Transformers 支持多种语言

L
scb10x
3,269
16
Cadet Tiny
Openrail
Cadet-Tiny是一个基于SODA数据集训练的超小型对话模型,专为边缘设备推理设计,体积仅为Cosmo-3B模型的2%左右。
对话系统
Transformers 英语

C
ToddGoldfarb
2,691
6
Roberta Base Chinese Extractive Qa
基于RoBERTa架构的中文抽取式问答模型,适用于从给定文本中提取答案的任务。
问答系统 中文
R
uer
2,694
98
智启未来,您的人工智能解决方案智库
简体中文