nlp

⭐ College.ai is an advanced AI-powered platform that provide a set of tools for students, job seekers, and professionals.

Automatically generate multi-language subtitles using AWS AI/ML services. Machine generated subtitles can be edited to improve accuracy and downstream tracks will automatically be regenerated based on the edits. Built on Media Insights Engine (https://github.com/awslabs/aws-media-insights-engine)

FTPipe and related pipeline model parallelism research.

SentimentArcs: a large ensemble of dozens of sentiment analysis models to analyze emotion in text over time

Baseline model for multimodal classification based on images and text. Text representation obtained from pretrained BERT base model and image representation obtained from VGG16 pretrained model.

An end-to-end image captioning project using a CNN encoder (ResNet-50) and LSTM decoder in PyTorch. Includes vocabulary building, preprocessing, training with BLEU evaluation, and inference. Generates natural language captions for images with saved metrics, model checkpoints, and visualization outputs.

Repository for the paper "MultiNERD: A Multilingual, Multi-Genre and Fine-Grained Dataset for Named Entity Recognition (and Disambiguation)" (NAACL 2022).

OpenLLM Monitor is a plug-and-play, real-time observability dashboard for monitoring and debugging LLM API calls across OpenAI, Ollama, OpenRouter, and more. Tracks tokens, latency, cost, retries, and lets you replay prompts — fully open-source and self-hostable.

Context-Based-Question-Answering

RATransformers 🐭- Make your transformer (like BERT, RoBERTa, GPT-2 and T5) Relation Aware!

Airflow Pipeline for Machine Learning

Read like a superhero: Turn PDFs into Bionic Reading format https://x.com/Benavent/status/1853508523638116689

End-to-end pipeline for seeing how LLMs actually process your prompts. Capture attention across every layer, render heatmaps and cooking curves, compare variants with evidence — not vibes.

Code for "Finetuning Pretrained Transformers into Variational Autoencoders"

StarListify is a Python package that classifies GitHub stars history into organized category lists based on user-defined criteria.

Papers of ASR, Tools of ASR

Use Large Language Models like OpenAI's GPT-3.5 for data annotation and model enhancement. This framework combines human expertise with LLMs, employs Iterative Active Learning for continuous improvement, and integrates CleanLab (Confident Learning) to ensure high-quality datasets and better model performance

A pipeline for Prompt-tuning

Experiment on NER task using Huggingface state-of-the-art Transformers Natural Language Models library

Official resource of the paper "Knowledge Enhanced Masked Language Model for Stance Detection", NAACL 2021