Generative AI Examples
Introduction
GenAIExamples are designed to give developers an easy entry into generative AI, featuring microservice-based samples that simplify the processes of deploying, testing, and scaling GenAI applications. All examples are fully compatible with both Docker and Kubernetes, supporting a wide range of hardware platforms such as Gaudi, Xeon, AMD EPYC CPUs, AMD Instinct GPUs, and other hardwares including NVIDIA GPUs, ensuring flexibility and efficiency for your GenAI adoption.
Architecture
GenAIComps is a service-based tool that includes microservice components such as llm, embedding, reranking, and so on. Using these components, various examples in GenAIExample can be constructed including ChatQnA, DocSum, etc.
GenAIInfra is part of the OPEA containerization and cloud-native suite and enables quick and efficient deployment of GenAIExamples in the cloud.
GenAIEval measures service performance metrics such as throughput, latency, and accuracy for GenAIExamples. This feature helps users compare performance across various hardware configurations easily.
Use Cases
Below are some highlighted GenAI use cases across various application scenarios:
| Scenario |
Use Case |
| Question Answering |
ChatQnA ✨: Chatbot with Retrieval Augmented Generation (RAG).
VisualQnA ✨: Visual Question-answering. |
| Image Generation |
Text2Image ✨: Text-to-image generation. |
| Content Summarization |
DocSum: Document Summarization Application. |
| Code Generation |
CodeGen: Gen-AI Powered Code Generator. |
| Information Retrieval |
DocIndexRetriever: Document Retrieval with Retrieval Augmented Generation (RAG). |
| Fine-tuning |
InstructionTuning: Application of Instruction Tuning. |
For the full list of the available use cases and their supported deployment type, please refer here.
Documentation
The GenAIExamples documentation contains a comprehensive guide on all available examples including architecture, deployment guides, and more. Information on GenAIComps, GenAIInfra, and GenAIEval can also be found there.
Getting Started
GenAIExamples offers flexible deployment options that cater to different user needs, enabling efficient use and deployment in various environments. Three primary methods are presently used to do this: Python startup, Docker Compose, and Kubernetes.
Users can choose the most suitable approach based on ease of setup, scalability needs, and the environment in which they are operating.
Deployment Guide
Deployment is based on released docker images by default - check docker image list for detailed information. You can also build your own images following instructions.
Prerequisite
-
For Docker Compose-based deployment, you should have docker compose installed. Refer to docker compose install for more information.
-
For Kubernetes-based deployment, you can use Helm or GMC-based deployment.
- You should have a kubernetes cluster ready for use. If not, you can refer to k8s install to deploy one.
- (Optional) You should have Helm (version >= 3.15) installed if you want to deploy with Helm Charts. Refer to the Helm Installation Guide for more information.
- (Optional) You should have GMC installed to your kubernetes cluster if you want to try with GMC. Refer to GMC install for more information.
-
Recommended Hardware Reference
Based on different deployment model sizes and performance requirements, you may choose different hardware platforms or cloud instances. Here are some of the reference platforms:
| Use Case |
Deployment model |
Reference Configuration |
Hardware access/instances |
| Xeon |
Intel/neural-chat-7b-v3-3 |
64 vCPUs, 365 GB disk, 100 GB RAM, and Ubuntu 24.04 |
Intel Tiber Developer Cloud |
| Gaudi |
Intel/neural-chat-7b-v3-3 |
1 or 2 Gaudi Cards, 16 vCPUs, 365 GB disk, 100 GB RAM, and Ubuntu 24.04 |
Intel Tiber Developer Cloud |
| Xeon (AWS) |
Intel/neural-chat-7b-v3-3 |
64 vCPUs, 100 GB disk, 64 GB RAM, and Ubuntu 24.04 |
AWS Cloud (e.g., c7i.16xlarge) |
| AMD EPYC |
meta-llama/Meta-Llama-3-8B-Instruct |
64 vCPUs, 100 GB disk, 256 GB RAM, and Ubuntu 24.04 |
Google Cloud Platform
Microsoft Azure
AWS |
| AMD Instinct |
meta-llama/Llama-3.1-405B |
GPU: 8× MI300X, 1536 GB vRAM, and Ubuntu 24.04 |
AMD Developer Cloud
Oracle Cloud Infrastructure
Microsoft Azure |
Deploy Examples
Note: Check for sample guides first for your use case. If it is not available, then refer to the table below:
Supported Examples
Check here for detailed information of supported examples, models, hardwares, etc.
Validated Configurations
Check here for the validated configurations of GenAIExamples, including hardware and software versions that have been tested for each release.
Contributing to OPEA
Welcome to the OPEA open-source community! We are thrilled to have you here and excited about the potential contributions you can bring to the OPEA platform. Whether you are fixing bugs, adding new GenAI components, improving documentation, or sharing your unique use cases, your contributions are invaluable.
Together, we can make OPEA the go-to platform for enterprise AI solutions. Let’s work together to push the boundaries of what’s possible and create a future where AI is accessible, efficient, and impactful for everyone.
Please check the Contributing guidelines for a detailed guide on how to contribute a GenAI component and all the ways you can contribute!
Thank you for being a part of this journey. We can’t wait to see what we can achieve together!
Additional Content
![dependabot[bot]](https://avatars.githubusercontent.com/in/29110?v=4)
评论