C_AIG_2412 - SAP Generative AI Developer Part 1

Which of the following architectures is primarily used by LLMs? Choose the correct answer. A neural network architecture knows as transformer. A mathematical calculator and function based on internet data known as modeler. Human text input combined with mathematical functions and calculator.

Which of the following are some of the key risks associated with using LLMs in the context SAP Business AI? There are three correct answers. Data Privacy Concerns. Misinterpretation of Data. Time-consuming calculations. Bias and Fairness.

How can businesses directly benefit from using the LLMs in the SAP AI Core? Choose the correct answer. They can only use predesigned and pretrained AI models. They can receive tailored responses or forecasts. They do not need to integrate the models into their applications. They do not need to input their unique data into the AI model.

What type of a setup is used for the LLMs in the context of the generative AI hub? Choose the correct answer. Hardware as a service (HaaS). Software as a service (SaaS). Platform as a service (PaaS). Infrastructure as a service (IaaS).

Which data framework is designed for developing data-aware applications powered by LLMs? Choose the correct answer. LangChain. LlamaIndex. SAP BTP. GPT Index.

What is a key focus area in the Validation Phase for data engineers and software architects? Choose the correct answer. Pricing and commercialization. Commercial aspect of the product. Technical evaluation and architectural guidelines. Real-world usage data and customer feedback.

What is the main focus of the Ideation phase in SAP's product development process? Choose the correct answer. Finalizing the pricing model for new products. Assessing the feasibility, desirability, and viability of LLMs. Identifying potential use cases for LLMs within SAP's ecosystem. Gathering customer feedback on developed use cases.

What is emphasized in the integration of frameworks, such as LangChain and LlamaIndex, in LLM application development using SAP's methodical approach? Choose the correct answer. Technical feat. Strategic orchestration. User experienced. Data intelligence.

How does the ToT technique differ from the CoT technique? Choose the correct answer. ToT is a linear, guiding the model through a single chain of reasoning steps to reach a conclusion. CoT involves presenting multiple related prompts in a linear chain. ToT allows the model to handle different strands of thought simultaneously, similar to branches on a tree. CoT creates a branching structure for exploring different paths in a conversation.

What is the primary goal of prompt engineering when working with LLMs? Choose the correct answer. To reduce the computational cost of training LLMs. To guide the model towards generating the desired output. To increase the size of the training dataset. To simplify the model's architecture.

Which of the following is the first step in the optimization journey for improving LLM performance? Choose the correct answer. Adding few-shot examples. Model fine-tuning. Integrating more context from a live knowledge base. Prompt engineering and evaluation.

What is the main purpose of the CoT and ToT techniques in relation to LLMs? Choose the correct answer. To limit the depth and quality of LLM responses. To improve the depth and quality of LLM responses. To restrict LLMs from exploring multiple lines of reasoning. To decrease the models' understanding of conversation context.

Which of the following methods is often the go-to method when relevance and reliability are crucial, and there is adequate context available for your LLM use case? Choose the correct answer. RAG approach. Model fine-tuning. Few-shot examples. Adding a fact-checking step.

What is the primary purpose of fine-tuning a Large Language Model (LLM)? Choose the correct answer. To introduce new knowledge to the model. For quick iterations on a new use case. To customize the model for specific tasks by continuing the training process on a new data set. To reduce the size of the model for easier deployment.

What is the primary goal of the Retrieval Augmented Generation (RAG) approach? Choose the correct answer. To reduce the size of large language models. To eliminate the need for any external knowledge base. To provide the model access to up-to-date, domain-specific information. To simplify the model's architecture.

What role do agents play in optimizing Large Language Model (LLM) performance? Choose the correct answer. They act as bridges between humans and LLMs, translating user input into prompts and interpreting outputs. They provide additional data sources and perform complex calculations. They are used to debug and inspect LLM output. They enhance the user experience by offering code completion.

How does the SAP HANA Cloud vector engine add more context to generative AI scenarios? Choose the correct answer. By processing and comparing vector data. By storing and analyzing vector embeddings. By integrating frameworks like LangChain. By providing a high-performance vector store.

What is the SAP HANA Cloud vector engine designed to handle? Choose the correct answer. Structured data. Unstructured vector data. Text data. Image data.

What do functions do in the context of LLMs? There are two correct answers. Translate user input into prompts for LLM. Provide access to external data sources. Improve the performance of LLM on specific use cases. Manage the interaction between user and LLM.

When is fine-tuning not ideal? Choose the correct answer. When introducing new knowledge to a model. When providing quick iterations on a new use case. When customizing outputs for specific types of inputs. When eliminating harmful content.

What is not a common capability explored at inference time? Choose the correct answer. Sentiment Analysis. Named entity recognition. Fine-tuning on domain-specific data. Language translation.

What is the impact of incremental updates to LLM training data? Choose the correct answer. Reduces the need for failover policies. Reduces the role of MLOps. Maintains performance as new data comes in. Increases the need for benchmarking.

Which metric helps avoid false positives? Choose the correct answer. F1 Score. Precision. Recall. Classification Accuracy.

Having recapped AI, which of these statements apply? Choose the correct answer. Machine Learning is a technology which encompasses AI, deep learning and generative AI. Deep Learning is a technology which encompasses AI, machine learning, and generative AI. AI is a technology which encompasses machine learning, deep learning and generative AI. Generative AI is a technology which encompasses AI, machine learning, and deep learning.

What are some examples of AI? There are two correct answers. Unlocking your cell phone using facial recognition. Receiving counted money at the cash terminal. Auto completion of sentences when writing an email. Getting a printed ticket from the ticket machine.

Why is it critical to continuously learn and adapt in the field of AI? There are two correct answers. AI is a hype topic everyone should know about. AI is already a major part in controlling every company. AI is one of the most transformative technologies of our times. AI is one of the most fast paced technology disruptions.

Which approaches in SAP BTP ensure that generative AI grounds and adapts to business context? There are three correct answers. Prompt engineering tools and templates. Integration to ChatGPT. Vector engine in SAP HANA Cloud. Fine-tuned models on specific scenarios. Adaptive search engine in SAP S/4HANA Cloud Public Edition.

Which of the following are key pillars of SAP's ethical AI framework announced in 2018? There are three correct answers. Foreseeableness and Simplicity. Transparency and Explainability. Human Agency and Oversight. Robustness and Performance. Addressing Bias and Discrimination.

What are the key requirement properties for SAP's AI in business? There are three correct answers. Relevant. Reliable. Replicable. Restrictive. Responsible.

Which of the following allow SAP Business AI to distinguish between embedded AI and extensibility approaches within its solution portfolio? There are three correct answers. By providing a clear legal framework. By offering all-in-one contracts. By offering embedded AI credits in all solutions. By collaborating with a wide ecosystem of partners.

What is special about Joule? Choose the correct answer. Joule is faster than any alternative. Joule understands business. Joule has a larger foundation than competitors. Joule works deterministically.

Which of the following characteristics apply to Joule? There are two correct answers. Joule only works with SAP data sources. Users interact with Joule in plain language. SAP Build Code is a Joule add-on for code generation. SAP Joule was first launched for SAP Ariba and Concur.

How could you describe AI Foundation? Choose the correct answer. A development environment. An engine for coding modules. A collaboration platform. A one-stop-shop for developers.

Which of the following are services offered by SAP’s AI Foundation? There are three correct answers. Pre-trained AI models for common tasks like Document Information Extraction. Instant access to a broad range of large language models. Simulated User Acceptance Testing and signoff. Tooling for prompt engineering and experimentation.

Which of the following are characteristics of a custom extension? There are two correct answers. It offers the highest level of customization and flexibility but requires significant development resources and expertise. Users interact with AI features within the familiar SAP environment, providing a cohesive user experience. The AI features are an integral part of the SAP Business application, enhancing its capabilities without requiring separate installations. Organizations create and implement AI models and applications from scratch, leveraging SAP development tools and frameworks.

Where can you find further information regarding specific use cases for Business AI? Choose the correct answer. On Interactive Value Journeys (IVjs). On Learning.SAP.com. On the SAP Help portal. On the SAP Roadmap Viewer.

Which of the following are the insights provided by the AI Foundation and the SAP ecosystem, including partnerships and investment areas? There are three correct answers. AI Foundation is a one-stop-shop for developers to bootstrap their own Business AI solutions. SAP partners with leading general-purpose AI vendors and large language model providers. SAP offers the resell of customer application built on SAP AI Foundation solutions. SAP offers a range of options for embarking on their AI journey.

What advantage does generative AI hub provide over traditional AI infrastructures? There are three correct answers. It allows trusted LLM and foundation model access, grounded on business and context data. It supports developing, deploying, and managing custom-built AI solutions. It provides access to multiple large language models. It saves your prompts or data automatically.

Which of the following interfaces does generative AI hub in SAP AI Launchpad offer? There are three correct answers. Chat. Document Editor. Prompt Editor. Prompt Management.

Which advantage provide open-weights LLMs? There are three correct answers. They reduce costs. They come with community support. They can only be used by large enterprises. They prevent vendor lock-in.

In which of the following contexts can generative AI revolutionize industries? There are two correct answers. Replacing human workers entirely in every sector. Automating intricate processes. Solely operating based on structured schema interactions. Offering valuable assistance for many tasks.

What is the primary role of SAP’s AI Foundation? There are two correct answers. To manage SAP's customer accounts. To extend SAP applications with AI and build custom AI solutions. To help developers orchestrate cutting-edge AI technology with business context. To train all employees on AI.

Which of the following tools does SAP AI Foundation provide to developers? There are two correct answers. Basic word processing applications. Tools to develop, deploy, and manage custom-built AI applications. Access to frontier AI models and generative AI foundation models. Financial auditing tools.

Where will the code for the ML pipelines be executed? Choose the correct answer. Cloud. AI Launchpad. Docker. Container.