Examples

• Native Client Integrations
  • Completions
    • OpenAI
      • Structured model outputs
    • Google GenAI
    • Amazon
  • Embeddings
    • OpenAI
    • Amazon
• Langchain Integration
  • Harmonized Model Initialization
  • LLM
  • Chat model
    • Structured model outputs
  • Embeddings
• SAP RPT-1 Models
  • Example of usage RPTClient for the regression task
  • Example of usage RPTClient for the classification task
  • Example of async usage of RPTClient
• Using New Models Before Official SDK Support
• Streaming
  • Native SDKs
    • OpenAI - ChatGPT
      • Structured model outputs
    • Google - GenAI
    • Anthropic - Claude
    • Amazon - Bedrock
  • Langchain
• Prompt Registry
  • Prerequisite
    • Provide the credentials to authenticate the client and establish a connection with the Prompt Registry API.
  • Prompt Template Management
    • Step 0: Initialize Client
      • Initialize the client to interact with the Prompt Registry.
    • Step 1: Create Prompt Templates
      • Define the Prompt Template configuration and post to Prompt Registry.
    • Step 2: Retrieve Prompt Templates
      • Retrieve the Prompt Template by ID.
    • Step 3: Modify the Prompt Template
      • We will add an input variable to the existing Prompt Template.
    • Step 4: Prompt Template History
      • Retrieve the history of Prompt Templates by scenario, name and version.
    • Step 5: Fill Prompt Template
      • Fill the variables in the Prompt Template.
  • Orchestration Config Management
    • Step 0: Initialize Client
      • Initialize the client to interact with the Prompt Registry.
    • Step 1: Create Orchestration Config
      • Define the Orchestration Config configuration and post to Prompt Registry.
    • Step 2: Retrieve Orchestration Config
      • Retrieve the Orchestration Config by ID.
      • Retrieve the Orchestration Configs by scenario, name and version..
    • Step 3: Orchestration Config History
      • Retrieve the history of Orchestration Config by scenario, name and version.
    • Export Orchestration Config
      • Export a design orchestration config in a declarative compatible yaml file.
• Orchestration Service [Deprecated]
  • Prerequisite
  • Authentication
  • Basic Orchestration Pipeline
    • Step 1: Define the Template and Default Input Values
    • Step 2: Define the LLM
    • Step 3: Create the Orchestration Configuration
    • Step 4: Run the Orchestration Request
      • Referencing Templates in the Prompt Registry
      • Overview of response_format Parameter Options
  • Understanding Deployment Resolution
    • Default Behavior
    • Direct Deployment Specification
    • Config-Based Specification
  • Optional Modules
    • Data Masking
      • Masking Types
      • Configuration Options
    • Content Filtering
  • Streaming
  • Tool Calling (Function Calling)
    • Defining Tools
      • Using the Python Decorator
      • Using the FunctionTool Class
      • Using a JSON Schema Dictionary
    • Synchronous Tool Call Workflow
    • Streaming Tool Calls
  • Using Images as Input
    • 1. Preparing Image Inputs
      • a) From a URL or Data URL
      • b) From a Local File
    • 2. Adding Images to a Prompt
  • Translation
  • Advanced Examples
    • Translation Service
    • Chatbot with Memory
    • Sentiment Analysis with Few Shot Learning
  • Async Support
• Orchestration Service V2 API
  • Prerequisite
  • Authentication
  • Basic Orchestration Pipeline
    • Step 1: Define the Template and Default Input Values
    • Step 2: Define the LLM
    • Step 3: Create the Orchestration Configuration
    • Step 4: Run the Orchestration Request
      • Referencing Templates in the Prompt Registry
      • Overview of response_format Parameter Options
  • Understanding Deployment Resolution
    • Default Behavior
    • Direct Deployment Specification
    • Config-Based Specification
  • Optional Modules
    • Data Masking
      • Masking Types
      • Configuration Options
    • Content Filtering
  • Streaming
  • Tool Calling (Function Calling)
    • Defining Tools
      • Using the Python Decorator
      • Using the FunctionTool Class
      • Using a JSON Schema Dictionary
    • Synchronous Tool Call Workflow
    • Streaming Tool Calls
  • Using Images as Input
    • 1. Preparing Image Inputs
      • a) From a URL or Data URL
      • b) From a Local File
    • 2. Adding Images to a Prompt
  • Translation
  • Advanced Examples
    • Translation Service
    • Chatbot with Memory
    • Sentiment Analysis with Few Shot Learning
  • Async Support
  • Embeddings
    • Basic Usage
    • Customizing Embedding Parameters
    • Batch Embeddings
    • Input Type Hints (Asymmetric Search)
    • Embeddings with Data Masking
      • Masking with Custom Entities and Allowlist
    • Async Embeddings
• Document Grounding
  • Prerequisites
    • Create a Vector knowledge base
  • Configuration of the Grounding Module
    • Create the configuration
      • Grounding configuration for searching SAP Help via elastic search
      • Grounding configuration for searching a custom data repository
      • One can also show the retrieved context from the grounding module, which is added to the prompt for improving the response.
      • Data Masking of the retrieved context
• Async examples
  • Async Amazon native
  • Async Function to Stream Amazon Model Response
  • Async Function to Test Amazon Titan Embedding
  • Async Google Genai native example
  • Async Google GenAI Chat Example
  • Async Google GenAI Stream Generate Content Example
  • Langchain examples
  • Async Chat Streaming Example
  • Chat Converse Model Example
  • Async Gemini Model Invocation Example
  • Async Gemini Streaming Example
• Generative AI Custom Evaluation
  • Setup
  • One Time Creation of Secrets
  • Helper function to list available models in the region
  • Helper function to see available System Defined Metrics
  • Defining the Evaluation Config
  • Helper function to generate Evaluation Config
  • Running the evaluate function:
  • Wait till each of the run is completed:
  • Debugging
  • Viewing the Aggregate Results
  • Viewing the Completion Response
  • Viewing the Metric Evaluation response