Access tutorials, blogs, open-source projects, and other resources for AI development with the ROCm™ software platform. This site provides an end-to-end journey for all AI developers who want to develop AI applications and optimize them on AMD GPUs.

AI development and deployment: Acquire resources, setup development environment, building and testing AI models, validating performance, and deploying the model into production environments with regular maintenance and support.

  1. Get Access to AMD GPUs
  2. ROCm and Container Setup
  3. Tutorials
  4. Performance Results
  5. Deployment and Orchestration

AMD GPUs for AI Development

AMD Instinct MI250

AMD Instinct™ GPUs

Large on-chip memory, state-of-the-art HBM memory and scale up/out capabilities ideal for generative AI in modern data center deployment. 

Image showing AMD Radeon™ PRO W7900 Graphics Card

AMD Radeon™ Graphics Cards 

Enable users to build GPU workstations suitable for local fine-tuning, and inference of relatively smaller models. 

Gain Access

Access GPUs, including cloud instances, OEM/ODM server systems, and Radeon PCIe® graphics cards.

AMD Cloud Access

AMD offers various cloud access programs to fit your needs. Experience the latest in GPU-accelerated cloud computing. Four cloud platforms are available to support your AI, HPC, and software development needs.

Learn About AMD Cloud Evaluation Programs

Featured Instinct OEM/ODM Partners

Build your own AI GPU clusters with AMD Instinct accelerator-powered solutions. Find solutions from our OEM/ODM/SI.

View Catalog
AMD Radeon Cards for AI Development

Learn more about Radeon cards built for AI development.

AMD Radeon™ PRO W7900 Professional Graphics
AMD Radeon™ PRO W7900 Dual Slot Professional Graphics
AMD Radeon™ PRO W7800 48GB Professional Graphics

ROCm Software and Container Setup

How to Tutorials for AI Workloads

Explore more topics to find relevant learning resources to help you stay updated with the latest technological advancements from AMD.

AI Inference

Deploy AI models faster with higher accuracy and achieve faster insights.

AI agent with MCPs using vLLM and PydanticAI
Speculative decoding with vLLM
Deploying Llama Stack
DeepSeek-R1 with SGLang
OCR with vision language models with vLLM
Building AI pipelines for voice assistants
Running Inference with Hugging Face Transformers
Running Hugging Face text generation inference (TGI)
Deploying models with vLLM
From chatbot and rap bot with vLLM
Constructing a RAG system: LlamaIndex and Ollama

For in-depth documentation for AI inference, visit the ROCm software documentation page.

Fine-tuning

Adapt a pre-trained model for specific tasks or use cases.

GRPO with Unsloth
Fine-tuning LLM with LlaMA Factory
Fine-tuning LLM with torchtune
Fine-tuning LLM with LoRA
Fine-tuning LLM with QLoRA
Fine-tuning a vision language model with the Hugging Face ecosystem (TRL)

For in-depth documentation for fine-tuning, visit the ROCm software documentation page.

Training

Pre-train a model on a large dataset to learn general features and patterns.

Custom diffusion model with PyTorch (DDIM)
Pretraining LLM with torchtitan
Pretraining OLMo model with PyTorch FSDP
Pretraining LLM with Megatron-LM
Training configuration with Megatron-LM

For in-depth documentation for fine-tuning, visit the ROCm software documentation page.

GPU Development and Optimization Tutorials

Kernel development and optimization with Triton
Profiling Llama-4 inference with vLLM
FP8 quantization with AMD Quark for vLLM

For in-depth documentation for optimization, visit the ROCm software documentation page. 

Watch quick video tutorials to get your questions answered.

How To Run DeepSeek on AMD GPUs with SGLang
Running an LLM on AMD GPUs
Deploying vLLM with AMD ROCm™ Software
How to install AMD ROCm software on Ubuntu

Performance Results

Find the latest performance on AMD GPUs. Performance numbers are frequently updated as a reference point to help users test observed performance.

See Performance Results

Deployment and Orchestration 

Streamline and automate the AI deployment, monitoring, and maintenance of machine learning models in production.

Explore Instinct™ Documentation

ROCm Documentation 

Comprehensive information about ROCm components, tools, and compilers, runtimes, and debugger.

Explore ROCm Documentation

Open-source Projects Using ROCm Software

Collaboration with ecosystem partners through CI/CD clusters and joint roadmap development.

See All

AMD Infinity Hub

Collection of advanced GPU software containers and deployment guides for AI and ML applications.

Explore more containers

Blogs

Stay updated on the latest tutorials, tips, and use cases for AI developers.

View All

Community and Support

Join various open-source forums, including our own ROCm discussion forum, to share knowledge, report issues, and get answers.

Featured Forums

Pytorch related questions and issues
vLLM related questions and issues
ROCm related questions
Ready to file bugs or PRs?

The ROCm™ Star Application Developer Certificate

Test your knowledge on developing, optimizing, and deploying AI and HPC applications on AMD GPUs using ROCm software and boost your career by earning an AMD ROCm Star Certificate.

Learn More