Reviewing AI and Generative AI Hub Concepts

In this lesson, we will revisit the essential concepts covered in the previous learning journey and prepare for the advanced techniques we will explore in this learning journey.

In the previous learning journey, you explored the fundamental concepts of Artificial Intelligence (AI) with a focus on SAP’s generative AI hub, equipping yourself with the knowledge to tackle advanced techniques and complex AI applications.

• AI Foundation on SAP BTP: The AI Foundation on SAP Business Technology Platform (BTP) provides a comprehensive set of tools and services designed to help businesses tackle various challenges. It includes capabilities for developing, deploying, and managing AI models, making it easier for organizations to integrate AI into their operations and drive innovation.
• Generative AI hub: The generative AI hub is a pivotal component within SAP AI Core, playing a crucial role in integrating AI capabilities into different applications. It serves as a central platform for accessing and deploying (LLMs), enabling businesses to apply the power of generative AI to solve complex problems and enhance their operations.
• Applications of Generative AI: Generative AI has diverse applications beyond traditional chatbots. It can be used to solve real-world problems in various business scenarios, such as automating customer service, generating content, and optimizing processes. Understanding these applications helps businesses identify opportunities to leverage generative AI for their specific needs.
• Developing Prompts: Developing effective prompts is essential for leveraging the power of generative AI. This involves creating prompts that are concise, specific, and relevant to the business problem at hand. The process includes initial setup, creating prompts with assigned values, and refining them to improve accuracy and relevance.
• Generative AI Hub SDK: The generative AI hub SDK provides practical tools for addressing business problems using LLMs. It includes examples of installing, configuring, and using the SDK, enabling businesses to develop and deploy AI solutions effectively. The SDK also supports advanced features like the orchestration service, which helps manage complex AI workflows.
• Evaluating Prompts: Evaluating prompts is crucial for ensuring their accuracy and relevance, especially when dealing with larger datasets. This involves implementing evaluation functions and using practical examples to assess the performance of prompts. Effective evaluation helps businesses refine their AI solutions and achieve better results.
• Advanced Prompt Engineering: Advanced prompt engineering techniques, such as few-shot prompting and metaprompting, are essential for improving the specificity, accuracy, and user experience of AI models. These techniques involve providing multiple examples to guide the model and crafting or refining prompts to generate more accurate and contextually relevant responses.
• Selecting Models: Selecting the right LLM is critical for achieving the desired performance and cost-effectiveness. This involves evaluating different models available in the generative AI hub, comparing their performance, and considering factors like specialization, flexibility, and reliability. Choosing the right model helps businesses tailor their AI solutions to specific needs and achieve optimal results.

This comprehensive recap provided a solid foundation for understanding advanced techniques like grounding, preparing us to explore the complexities of advanced AI applications.

The Model Library in the Generative AI Hub is an invaluable resource for selecting the right model. It provides comprehensive information on available models to aid in decision-making.

• Catalog Mode: Explore all available models and their metadata. Use filters to refine your selection or search for a model by name. Each model's card offers detailed information, including data input types, cost, and metrics.

• Leaderboard Mode: Access model scores across various benchmarks. Apply filters to narrow down your options, search by name, or reorder the list based on specific benchmarks. Deprecation notices are also available.

• Chart Mode: View model scores in a chart format for easy comparison.

• Model Cards: Navigate through tabs to see detailed information and the status of deployments containing the model. Refresh to get the latest updates.

See more details here.

In this learning journey, we will build on the foundational knowledge from the previous learning journey and delve into advanced AI techniques. The upcoming units and lessons will cover:

• Orchestration: Exploring how the orchestration service can streamline AI workflows.
• Vector Embeddings: Learning about vector embeddings and their applications in AI.
• Embedding Models: Identifying embedded models to optimize AI responses.
• Retrieval Augmented Generation (RAG): This technique involves using relevant documents to provide accurate, contextually appropriate responses. Document repositories, such as Microsoft SharePoint, are indexed, allowing the AI to retrieve and use specific data from these documents.
• Document Grounding: Understanding and implementing document grounding to enhance AI applications. Grounding is also known as RAG, which is an important tool for specializing LLMs on domain knowledge without the need of retraining.

This lesson is a recap of the foundational concepts of AI and the generative AI hub. This knowledge is essential as we move forward in developing and refining advanced AI techniques for various business applications.