Reasoning-BERT-CCNews Open-Source Model - Realize Sentence and Paragraph Mapping, Support Tasks Such as Semantic Search

Reasoning Bert Ccnews

Developed by bwang0911

This is a fine-tuned BERT-based sentence transformer model for mapping sentences and paragraphs into a 768-dimensional vector space, supporting tasks such as semantic text similarity and semantic search.

Text Embedding

Safetensors

#Semantic Text Similarity #Reasoning-Guided Ranking #News Content Analysis

Downloads 13

Release Time : 3/28/2025

Model Overview

This model is fine-tuned based on google-bert/bert-base-uncased, trained on the reason_unfiltered dataset, primarily used for natural language processing tasks such as semantic text similarity calculation and information retrieval.

Model Features

High-Quality Sentence Embeddings

Maps sentences and paragraphs into a 768-dimensional dense vector space, preserving semantic information.

Reasoning-Guided Ranking Loss Training

Trained using a reasoning-guided ranking loss function to optimize semantic similarity judgment.

Multi-Task Performance Optimization

Performs well on multiple information retrieval tasks, including datasets such as nfcorpus and trec-covid.

Model Capabilities

Semantic text similarity calculation

Semantic search

Paraphrase mining

Text classification

Clustering analysis

Use Cases

Information Retrieval

Document Similarity Search

Finding semantically similar documents in large-scale document collections

Achieved an accuracy@10 of 0.5975 on the nfcorpus dataset.

Question Answering System

Matching the semantic similarity between questions and candidate answers

Achieved an accuracy@1 of 0.7256 on the quora dataset.

News Analysis

News Content Similarity Analysis

Analyzing the semantic relevance between different news articles

Trained on the CCNews dataset, suitable for news text processing.

🚀 Sentence Transformer for Sentence Similarity

This project is a sentence transformer model based on google-bert/bert-base-uncased, designed for tasks such as sentence similarity and feature extraction. It offers valuable insights and performance in information retrieval tasks.

🚀 Quick Start

This model can be used directly through the sentence-transformers library. You can load the model and start performing sentence similarity tasks.

✨ Features

Sentence Similarity: Effectively calculate the similarity between sentences, useful for information retrieval and text matching.
Feature Extraction: Extract high - quality features from sentences for downstream tasks.
Generated from Trainer: The model is generated through a training process, ensuring its performance and generalization ability.

📦 Installation

Since this model is based on the sentence-transformers library, you can install it using the following command:

pip install sentence-transformers

💻 Usage Examples

Basic Usage

from sentence_transformers import SentenceTransformer
model = SentenceTransformer('your_model_name')
sentences = ["This is an example sentence", "Each sentence is converted"]
embeddings = model.encode(sentences)
print(embeddings)

Advanced Usage

# In a more complex scenario, you can use the model for information retrieval
from sentence_transformers import SentenceTransformer, util
model = SentenceTransformer('your_model_name')
query = "What is the meaning of life?"
corpus = ["Life is a journey", "The meaning of life is to find happiness"]
query_embedding = model.encode(query)
corpus_embeddings = model.encode(corpus)
hits = util.semantic_search(query_embedding, corpus_embeddings, top_k=1)
print(hits)

📚 Documentation

Datasets

bwang0911/reasoning_pairs_filtered_w_reason_ccnews

Pipeline Tag

sentence-similarity

Library Name

sentence-transformers

Metrics

Property	Details
Metrics	cosine_accuracy@1, cosine_accuracy@3, cosine_accuracy@5, cosine_accuracy@10, cosine_precision@1, cosine_precision@3, cosine_precision@5, cosine_precision@10, cosine_recall@1, cosine_recall@3, cosine_recall@5, cosine_recall@10, cosine_ndcg@10, cosine_mrr@10, cosine_map@100

Model Index

Name: SentenceTransformer based on google-bert/bert-base-uncased

Results:

Task:
- Type: information - retrieval
- Name: Information Retrieval
Dataset:
- Name: mteb/nfcorpus
- Type: mteb/nfcorpus

Metrics:

Property	Details
cosine_accuracy@1	0.3126934984520124
cosine_accuracy@3	0.47678018575851394
cosine_accuracy@5	0.5325077399380805
cosine_accuracy@10	0.5975232198142415
cosine_precision@1	0.3126934984520124
cosine_precision@3	0.2549019607843137
cosine_precision@5	0.20990712074303408
cosine_precision@10	0.16563467492260062

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご