These projects develop and evaluate AI/NLP tools for the Armenian language, including LLMs, benchmarks, OCR systems, text classification, and language preservation applications.
An Armenian OCR pipeline combining CRAFT for text detection and a deep learning model for text recognition, specifica...
An open-source OCR model trained for Armenian text recognition, supporting Classical, Western, and Eastern Armenian. ...
Nomikos is a comprehensive open-source platform for transcribing historical manuscripts, specifically designed for sc...
ArmBench-LLM is a specialized evaluation framework for benchmarking large language models on Armenian language tasks....
A Western Armenian language model project using transformer architecture, implemented in Jupyter notebooks. Includes ...
A project exploring Armenian language autocomplete using multiple NLP approaches including Word2Vec, LSTM, BERT trans...
This repository contains the complete reproduction code for the research paper "From Zero to Hero: An Open LLM Ecosys...
AzgIntel is an AI-powered NLP pipeline for Armenian text classification using a BERT-based multilingual model. It pro...
This repository provides two open-source OCR text recognition models (tiny and small) specifically trained for Classi...
A project to create a bilingual Western Armenian-English large language model using QLoRA fine-tuning on Qwen 2.5 1.5...
An AI-powered system for analyzing scanned Armenian newspapers using computer vision (YOLO for section detection, Ope...
A machine learning project for detecting AI-generated text in Armenian. It includes a custom dataset, fine-tuned tran...
ArmGPT is a student-friendly implementation of a GPT language model specifically trained on Armenian text. It's built...
OCArm is a full-stack OCR project for handwritten Armenian pages, using CRAFT word detection and a fine-tuned ResNet-...
A project to build a Large Language Model (LLM) specifically for the Armenian language from scratch using PyTorch and...
A Python package that adds Armenian language support to EasyOCR by providing a trained recognition model, weight cach...
A training pipeline for creating an Armenian OCR text recognition model compatible with EasyOCR. It generates synthet...