azizbarank

Czech-T5-Base-Model

This is the t5 base model for the Czech that is based on the smaller version of the google/mt5-base model. To make this model, I retained only the Czech and some of the English embeddings from the original multilingual model.

Amsterdam-Airbnb-Analysis

Airbnb data analysis of Amsterdam using Power BI.

California-Infectious-Diseases-Analysis

Data analysis of selected communicable infectious diseases reported for California residents between 2001-2022.

Chicago-Crime-Rate-Analysis

Analysis of trends in crime rates in Chicago between 2001-2024

Connecticut-Real-Estate-Analysis

Analysis of real estate sales in the state of Connecticut between 2001-2022

distilroberta-base-sst-2-distilled

Using Task Specific Knowledge Distillation to obtain DistilRoBERTa model fine-tuned on SST-2 part of the GLUE dataset for sentiment analysis.

Fine-tuning-MarianMT-for-English-Vietnamese-Translation

This project uses and fine-tunes MarianMT transformer for English to Vietnamese translation task

Text-Translation-and-Summarization-App-from-Turkish-to-English

This project applies Gradio and HuggingFace Transformers to make a demo app applying translation and summarization tasks in series to a Turkish text for its English equivalent output.

Topic-Modelling-with-spaCy-Gensim-and-scikit-learn

This project uses spaCy, Gensim, and scikit-learn for topic modeling on the NeurIPS (NIPS) Papers dataset. While spaCy is used for text preprocessing, Gensim and scikit-learn are used for using Latent Dirichlet Allocation (LDA) topic modelling algorithm.

Toxic-Comment-Detector

This project applies classification models with the aim of automating the detection of toxic comments on social media. After choosing the model with the best performance, HuggingFace + Streamlit are applied to make the web app.

azizbarank

azizbarank.github.io

My personal website where I share my NLP experience through blog posts.

Financial-Sentiment-Analysis-with-Machine-Learning-LSTM-and-BERT-Transformer

This project applies three main methods to make sentiment analysis on financial data: Machine Learning, LSTM using TensorFlow with Keras API, and BERT Transformer using the "simpletransformers" library.

Turkish-Sentiment-Analyser

This project fine-tunes the distilled Turkish BERT model on a review dataset for doing sentiment analysis. After the fine-tuning, Hugging Face Spaces and Streamlit are used to deploy the final model for other people to use it without any code.

Dutch-GPT-2-Text-Generation

Using the Dutch version of the GPT-2 model that was obtained by retraining lexical embeddings and then additional training to make a text generation app.

French-Zero-Shot-Text-Classification

Using CamemBERT and multilingual XLM-R models to make a French zero-shot text classification web application.

Kaggle-AG-News-Machine-Learning-Classifiers

This project uses scikit-learn's built-in different Machine Learning (ML) classifiers and the "Perceptron" classifier to categorize the text data of the third version of AG News dataset on Kaggle.

NLP-Tutorials

This repository contains the posts I upload with the aim of sharing my NLP experience

Text-Summarization-with-T5-Pegasus-and-Bart-Transformers

This project uses T5, Pegasus and Bart transformers with HuggingFace for text summarization applied on a news dataset on Kaggle.