R

Rum2m100 1.2B

Developed by ai-forever
A Russian spelling check model trained based on M2M100-1.2B, capable of correcting spelling and typing errors
Downloads 407
Release Time : 7/26/2023

Model Overview

This model corrects spelling and typing errors by normalizing all words in the text to standard Russian forms, trained on an extensive dataset containing artificial errors

Model Features

High-quality spelling correction
Outperforms ChatGPT series models and open-source spell checkers on multiple test datasets
Multi-domain applicability
Performs well in various domains including general text, medical text, and technical text
Trained on synthetic errors
Uses the SAGE library to automatically generate errors in training data, covering a wide range of error types

Model Capabilities

Russian spelling check
Typing error correction
Text normalization

Use Cases

Text processing
Social media text correction
Corrects spelling and typing errors in social media texts
F1 score of 50.1 on the RUSpellRU dataset
Medical text processing
Corrects spelling errors of professional terms in medical histories
F1 score of 60.6 on the MedSpellChecker dataset
Technical document processing
Corrects spelling errors in technical documents and code comments
F1 score of 43.5 on the GitHubTypoCorpusRu dataset
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase