MachineLearning

This repo contains a few machine learning projects using LLMs, Deep Learning, and Matrix decompositions (SVD/PCA)

LLMs

Chatbots

• Simple Chatbot Powered by Mistral, Gemma, and Zephyr

  • In this project we look at creating a simple chatbot using Streamlit and pulling models from Hugging Face for inference.
    We use the Mistral 7B model from Mistral AI, the Gemma models from Google, and the Zephyr model which is a fine tuned Mistral model by the team at Hugging Face.
    • You can read more about the project here.
    • You can try try the interactive demo here.
      Note: The Gemma models are currently being updated for inference and may not work at times. Hugging Face periodically updates and models may be unavailable at times.
    
    

Deep Learning

Image Recognition & Classification

• ResNet50

  • Here we use the ResNet50 model in keras to perform image recognition. Notebook code here.
    

• VGG16

  • Here we use the VGG16 model in keras to perform image recognition. Notebook code here.
    

• Simple Classifier From Scratch

  • Here we build a deep neural network to classify hand written digits from the MNIST dataset. Notebook code here.
    

• PCA / SVD Image Decomposition

  • Here we use a combination of PCA and SVMs to classify famous faces. Notebook code here.