AI Developer Hub

Welcome to the AI Developer Hub repository! This project contains Jupyter Notebook tutorials and guides for training, fine-tuning, and inference using popular machine learning frameworks on AMD GPUs.

Current Notebooks

Category	Title	GitHub Link	AMD Tutorial Page
Inference Tutorials	ChatQnA vLLM deployment and performance evaluation	GitHub	AMD Tutorial
	Text-to-video generation with ComfyUI on Radeon GPU	GitHub	AMD Tutorial
	DeepSeek Janus Pro on CPU or GPU	GitHub	AMD Tutorial
	DeepSeek-R1 with vLLM V1	GitHub	AMD Tutorial
	AI agent with MCPs using vLLM and PydanticAI	GitHub	AMD Tutorial
	Multi-agents with Google ADK and A2A protocol	GitHub	AMD Tutorial
	Hugging Face Transformers	GitHub	AMD Tutorial
	Deploying with vLLM	GitHub	AMD Tutorial
	From chatbot to rap bot with vLLM	GitHub	AMD Tutorial
	RAG with LlamaIndex and Ollama	GitHub	AMD Tutorial
	OCR with vision-language models with vLLM	GitHub	AMD Tutorial
	Building AI pipelines for voice assistants	GitHub	AMD Tutorial
	Speculative decoding with vLLM	GitHub	AMD Tutorial
	DeepSeekR1 with SGLang and example applications	GitHub	AMD Tutorial
	PD disaggregation with SGLang	GitHub	AMD Tutorial
	Accelerating DeepSeek-V3 inference using multi-token prediction in SGLang	GitHub	AMD Tutorial
Fine-Tuning Tutorials	Customize Qwen-Image with DiffSynth-Studio	GitHub	AMD Tutorial
	VLM with PEFT	GitHub	AMD Tutorial
	Llama-3.1 8B with torchtune	GitHub	AMD Tutorial
	Llama-3.1 8B with Llama Factory	GitHub	AMD Tutorial
	GRPO with Unsloth	GitHub	AMD Tutorial
Pretraining Tutorials	Training configuration with Megatron-LM	GitHub	AMD Tutorial
	LLM with Megatron-LM	GitHub	AMD Tutorial
	Llama-3.1 8B with torchtitan	GitHub	AMD Tutorial
	Draft model training with SGLang SpecForge	GitHub	AMD Tutorial
	Custom diffusion model	GitHub	AMD Tutorial
	Pretraining with TorchTitan	GitHub	AMD Tutorial
	Training a model with Primus	GitHub	AMD Tutorial
GPU Development and Optimization Tutorials	Accelerating Quark MXFP4 quantization for vLLM	GitHub	AMD Tutorial
	Kernel development and optimizations with Triton	GitHub	AMD Tutorial
	Profiling Llama-4 inference with vLLM	GitHub	AMD Tutorial
	FP8 quantization with AMD Quark for vLLM	GitHub	AMD Tutorial
About	Licensing and Support Information	GitHub	AMD Tutorial

Repository Structure

The tutorials are organized into four main categories:

• Fine-Tuning: Examples and guides for fine-tuning machine learning models.
• Pretraining: Tutorials on pretraining models from scratch.
• Inference: Resources for running inference with trained models.
• GPU development and optimization: Resources for optimizing AI compute and kernel development on GPUs.

Directory Layout

github_repo/
├── docs/                          # Documentation for the tutorials
│   ├── index.md                   # Main documentation index
│   ├── fine_tune.md               # Fine-tuning tutorials index
│   ├── pretrain.md                # Pretraining tutorials index
│   ├── inference.md               # Inference tutorials index
│   └── notebooks/                 # Jupyter notebooks organized by category
│       ├── gpu_dev_optimize/      # GPU development and optimization notebooks
│       ├── fine_tune/             # Fine-tuning notebooks
│       ├── pretrain/              # Pretraining notebooks
│       └── inference/             # Inference notebooks