Muhammad Ali Shafique

I am a Ph.D. student at Kansas State University. As a graduate research assistant at the ISCAAS Lab, my research involves designing efficient machine learning models, specifically large language models (text, vision) and classical machine learning models, for resource-constrained systems.

Top skills:
Large Language Models (text, vision), Machine Learning, MLOps, and Parallel Computing

Email  /  Linkedin  /  GitHub  /  Medium

Lightweight Large Language Models with Teacher-Student Knowledge Distillation

The goal of this research is to distill the rich representations and language understanding capabilities of the teacher model into the student model using knowledge distillation and RLHF. It enables student model to achieve comparable performance with reduced computational resources.

Deep Learning Performance Characterization on GPUs for Various Quantization Frameworks

This study has employed various performance metrics to identify the key factors in model architectures that impact quantization performance...

End to End LLM Chatbot Using Langchain and OpenAI in Finance Domain.
Live Demo

A repository that provides a thorough collection of frameworks and methods used for evaluating LLMs.
GitHub  |  Medium

Finetuning of LLM (model: T5-3b) on single GPU using QLoRA for summarization task.
GitHub  |  Medium

Prompt engineering techniques and best practices for desirable and precise responses.
GitHub  |  Medium

PyTorch training optimization:
5× throughput with GPU profiling and memory analysis.
GitHub  |  Medium

Efficient Inference optimizations and benchmarking of the model using post-training quantization.
GitHub  |  Medium

Last Updated: August 2024
Credits for the template to Jon Barron.