T

Titulm Llama 3.2 3b V2.0

Developed by hishab
A large Bengali language model optimized based on the Llama-3.2-3B architecture, with 42K Bengali tokens extended and fine-tuned, performing excellently in Bengali understanding and generation tasks.
Downloads 2,669
Release Time : 10/16/2024

Model Overview

This model is an autoregressive language model optimized for Bengali, mainly used for Bengali text generation and understanding tasks, and also supports English.

Model Features

Bengali optimization
Extended 42K Bengali tokens, significantly improving the processing ability for Bengali
Multi-source training data
Trained with 268GB of Bengali data from various sources, including web documents, books, and translated texts
High-performance evaluation results
Outperforms the original Llama-3.2-3B model in multiple metrics in Bengali benchmark tests
Long context support
Supports a context length of 4096 tokens

Model Capabilities

Bengali text generation
Bengali language understanding
English text processing
Instruction fine-tuning tasks

Use Cases

Natural language processing
Bengali question-answering system
Used to build Bengali question-answering applications
Achieved an accuracy of 0.59 on the BoolQ Bangla dataset
Bengali content generation
Automatically generate Bengali articles, stories, and other content
Education
Language learning assistance
Help learners understand and generate Bengali
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase