Back to Projects

Translatica

A fine-tuned Llama-3.2-3B-Instruct model using QLoRA/LoRA for multilingual sentiment analysis. Features Bengali language support and Langchain integration for Sequential Chaining and Inference.

Date
Views
500

Translatica

Innovative project showcasing modern web technologies

Overview

Translatica is an advanced machine learning model fine-tuned specifically for multilingual sentiment analysis. It leverages the powerful Llama-3.2-3B model with QLoRA optimization techniques to achieve excellent performance on consumer-grade hardware. The model includes special support for Bengali language and integrates seamlessly with LangChain for building complex inference chains.

Tech Stack

LangChainUnslothLlama-3.2-3B-InstructQLoRALoRA

Features

  • Multilingual sentiment analysis
  • Bengali language support
  • Optimized with QLoRA
  • LangChain integration
  • Inference chaining support