EuroBERT-210m-Quality Open Source Model - Automatically evaluate the quality of language text data, with two options available

Eurobert 210m Quality

Developed by TempestTeam

A model for automatically evaluating the quality of text data in both natural and programming languages, offering both unified and independent model solutions.

Text Classification

Transformers

Supports Multiple LanguagesOpen Source License:Apache-2.0 #Multilingual Quality Assessment #Code Quality Detection #Harmful Content Filtering

Downloads 26

Release Time : 3/18/2025

Model Overview

This model automatically assesses text data quality through a scoring system, supporting natural languages (French, English, Spanish) and programming languages (Python, Java, JavaScript, C/C++). It identifies harmful content and classifies quality levels.

Model Features

Multilingual Support

Supports quality evaluation for both natural languages (French, English, Spanish) and programming languages (Python, Java, JavaScript, C/C++).

Dual-Model Solution

Offers both unified and independent model solutions, allowing users to choose the most suitable evaluation approach based on their needs.

Harmful Content Detection

Efficiently identifies harmful content with F1 scores of 0.93 (natural language) and 0.79 (programming language).

Quality Level Classification

Classifies text data into four levels: harmful content, low quality, medium quality, and high quality, facilitating subsequent processing.

Model Capabilities

Natural Language Text Quality Evaluation

Programming Language Code Quality Evaluation

Harmful Content Detection

Quality Level Classification

Use Cases

NLP Pipeline

Automatic Text Corpus Validation

Automatically validates the quality of text corpora in NLP pipelines to enhance model training effectiveness.

Accuracy ~82% (natural language)

Community Content Management

Forum Content Quality Evaluation

Automatically evaluates the quality of forum or Stack Overflow content to assist in content management.

Harmful content detection F1 score 0.93 (natural language)

Code Generation

Code Quality Evaluation

Automatically evaluates the quality of generated code in code generation workflows to ensure usability.

Accuracy ~63% (programming language)

🚀 Automatic Evaluation Models for Textual Data Quality (NL & CL)

Automatically assess the quality of textual data using a clear and intuitive scale, suitable for both natural language (NL) and code language (CL).

🚀 Quick Start

This project offers two distinct approaches to automatically evaluate the quality of textual data:

A unified model that handles both NL and CL jointly: EuroBERT-210m-Quality
A dual - model approach that treats NL and CL separately:
- EuroBERT-210m-Quality-NL for natural language
- EuroBERT-210m-Quality-CL for code language.

✨ Features

Classification Categories

Harmful: Harmful data, potentially incorrect or dangerous.
Low: Low - quality data with major issues.
Medium: Medium quality, improvable but acceptable.
High: Good to very good quality data, ready for use without reservation.

Supported Languages

Natural Language: French 🇫🇷, English 🇬🇧, Spanish 🇪🇸
Code Language: Python 🐍, Java ☕, JavaScript 📜, C/C++ ⚙️

📚 Documentation

Performance

f1 - score: Unified Model (NL + CL)

Category	Global (NL + CL)	NL	CL
Harmful	0.86	0.93	0.79
Low	0.62	0.81	0.40
Medium	0.63	0.78	0.50
High	0.77	0.81	0.74
Accuracy	0.73	0.83	0.62

f1 - score: Separate Models

Category	Global (NL + CL)	NL	CL
Harmful	0.83	0.93	0.72
Low	0.64	0.76	0.53
Medium	0.63	0.76	0.52
High	0.79	0.81	0.76
Accuracy	0.73	0.82	0.63

Key Performance Metrics

Unified Model (NL + CL):
- Overall accuracy: ~73%
- High reliability on harmful data (f1 - score: 0.86)
Separate Models:
- Natural Language (NL): ~82% accuracy
  - Excellent performance on harmful data (f1 - score: 0.93)
- Code Language (CL): ~63% accuracy
  - Good detection of harmful data (f1 - score: 0.72)

Training Dataset

Public dataset available: [TempestTeam/dataset - quality](https://huggingface.co/datasets/TempestTeam/dataset - quality)

Common Use Cases

Automatic validation of text corpora before integration into NLP or code generation pipelines.
Quality assessment of community contributions (forums, Stack Overflow, GitHub).
Automated pre - processing to enhance NLP or code generation system performance.

Recommendations

⚠️ Important Note

For specialized contexts, use the separate NL and CL models for optimal results.

💡 Usage Tip

The unified model is suitable for quick assessments when the data context is unknown or mixed.

Citation

Please cite or link back to this model on Hugging Face Hub if used in your projects.

📄 License

This project is licensed under the Apache 2.0 license.

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