T

Turkish Colbert

Developed by ytu-ce-cosmos
Turkish paragraph retrieval model based on ColBERT architecture, fine-tuned on the Turkish-translated MS MARCO dataset
Downloads 1,724
Release Time : 12/3/2024

Model Overview

This is a Turkish paragraph retrieval model based on the ColBERT architecture, specifically designed for Turkish paragraph retrieval tasks. The model was fine-tuned on 500,000 triplets from the Turkish-translated MS MARCO dataset.

Model Features

Turkish optimization
Paragraph retrieval model specifically optimized for Turkish, fine-tuned on a Turkish foundation BERT model
Efficient retrieval
Utilizes ColBERT architecture to provide efficient paragraph retrieval capabilities
Case handling
Provides special case handling solutions for Turkish-specific 'I' character issues

Model Capabilities

Turkish paragraph retrieval
Sentence similarity calculation
Document indexing and search

Use Cases

Information retrieval
Scientific literature retrieval
Retrieve relevant information from scientific literature databases
Achieved 48.38 R@1 recall rate on Scifact-tr dataset
Encyclopedia knowledge retrieval
Retrieve relevant information from encyclopedia knowledge bases
Achieved 31.21 R@1 recall rate on WikiRAG-TR dataset
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase