Subham Kundu's repositories
KnowledgeDistillationOfLLM
A hand's on approach to learn how to do knowledge distillation on LLMs
LLMClients
This is my repository for storing utility snippets for different LLM clients I use in my projects
LLMSummaEvaluation
This project is meant for evaluating the summarization capabilities of three Large Language Models (T5-Small, T5-Large and GPT2). The dataset used in CNN_Daily Mail dataset
FineTuningLlama2
This is a repository where I have fine-tuned an Llama2 on an open source dataset from Kaggle.
TransformerFromScratch
This is an implementation of transformer for translating english to italian from scratch in pytorch
agentkit
Starter-kit to build constrained agents with Nextjs, FastAPI and Langchain
AsyncLLMInferenceAWS
In this project, I have created an async inference in SageMaker that queues incoming requests and processes them asynchronously.
aws-codepipeline-lab-repo
Cloud lab resources for the " Working with AWS CodeBuild, CodeDeploy, and CodePipeline" lab on Educative.
cenrax.portfolio
This is the codebase for my portfolio.
codellama
Inference code for CodeLlama models
companion
Generative-AI-Powered Foreign-Language Private Tutor
dinov2
PyTorch code and models for the DINOv2 self-supervised learning method.
FinetuningTextToCypher
This is a demo project where I have finetuned Llama3 8b model quantized in 4bit
giskard
🐢 The testing framework for ML models, from tabular to LLMs
LLM-Finetuning
LLM Finetuning with peft
MetagptCompletedProjects
This repo contains all the projects generated by metagpt but the finished code can be found here
MultimodalRag
This is a side project where I have built a multimodel RAG which scrapes a html page and embeds the image and text in a vector database(qdrant) and then answers the queries about that webpage
nextjs-starter
This NextJS starter project will help developers quickly stand-up a website using NextJS and supporting technologies.
OLMo
Modeling, training, eval, and inference code for OLMo
sec-insights
A real world full-stack application using LlamaIndex