Michael Hu

alpha_zero. A PyTorch implementation of DeepMind's AlphaZero agent to play Go and Gomoku board games

deep_rl_zoo. A collection of Deep Reinforcement Learning algorithms implemented with PyTorch to solve Atari games and classic control tasks like CartPole, LunarLander, and MountainCar.

InstructLLaMA. Implements pre-training, supervised fine-tuning (SFT), and reinforcement learning from human feedback (RLHF), to train and fine-tune the LLaMA2 model to follow human instructions, similar to InstructGPT or ChatGPT, but on a much smaller scale.

Llama3-FunctionCalling. Fine-tune Llama3 model to support function calling

muzero. A PyTorch implementation of DeepMind's MuZero agent

SAP-UI5-Development-Re-Introduction. This is the official source code for Udemy course SAP UI5 Development Re-Introduction

QLoRA-LLM. A simple custom QLoRA implementation for fine-tuning a language model (LLM) with basic tools such as PyTorch and Bitsandbytes, completely decoupled from Hugging Face.

miniGPT. Try to implement pre-training and fine-tuning GPT-2 model for research and education purpose.

DPO-LLaMA. A clean implementation of direct preference optimization (DPO) to train the LLaMA 2 model to align with human preferences.

RAG-LLaMA. A clean and simple implementation of Retrieval Augmented Generation (RAG) to enhanced LLaMA chat model to answer questions from a private knowledge base. We use Tesla user manuals to build the knowledge base, and use open-source embedding and Cross-Encoders reranking models from Sentence Transformers in this project.

MM-LLaMA. Bring multimodality to the LLaMA model by leveraging ImageBind as the modal encoder. This project supports vision input (both images and short videos) to the LLaMA model, with text output generated by LLaMA.

triton_llm_inference. Study on how to build a simple self-hosting LLM interference solution