An educational app powered by Gemini, a large language model provides 5 components a chatbot for real-time Q&A,an image & text question answerer,a general QA platform, a tool to generate MCQs with verified answers, and a system to ask questions about uploaded PDFs.
Fine-tuned Llama-2-7b-chat-hf using LoRA/qLoRA on a 701-row Bhagavad Gita dataset in Colab, achieving a loss of 1.25 for spiritually aligned, accurate responses in an interactive chat application.
An AI-powered medical chatbot using the Llama-2-7B-Chat model for precise clinical responses. Integrates Chroma DB and all-MiniLM-L6-v2 embeddings trained on medical literature, including texts like Clinical Emergency Medicine and Gale Encyclopedia. Accurate, fast, and reliable for healthcare queries.
A platform for sharing diverse knowledge often faces the challenge of managing questions that are worded in a similar way. In order to enhance user experience and easy access to high quality answers, we employ Natural Language Processing (NLP) tools to identify and remove similar questions. Specifically, we represent the text using Term Frequency-Inverse Document Frequency (TF-IDF) and Bag of Words (BoW), while a Random Forest classifier is used to detect duplicates accurately. This makes it easier for users to find relevant information quickly and effortlessly.
The Plant Disease Detection utilizes CNN architectures, including AlexNet, VGG-16, VGG-19, ResNet, DenseNet, EfficientNet, and ConvNextLarge, leveraging deep learning and transfer learning to identify plant diseases from leaf images, fostering sustainable agriculture and food security.
The Emotion Detection project leverages CNN architectures like Custom CNN, VGG16, and ResNet to identify emotions from facial images. This deep learning solution aims to enhance human-computer interaction and emotional intelligence, with applications in virtual assistants and mental health support systems.
The Sentiment Analysis project leverages NLP techniques such as Word2Vec, n-gram, and Bag of Words to accurately classify text sentiment. It includes implementation notebooks and a Streamlit web application for real-time sentiment detection.