Generative AI 2025

You need to build an LLM application using Oracle Database 23ai as the vector store and OCI Generative Al service to embed data and generate responses. What could be your approach?. Use LangChain classes to embed data outside the database and generate response. Use Select Al. Use LangChain Expression Language (LCEL). Use DB Utils to generate embeddings and generate response using SQL.

How are fine-tuned customer models stored to enable strong data privacy and security in OCI Generative Al service?. Stored in OCI Object Storage and encrypted by default. Stored in an unencrypted form in OCI Object Storage. Stored in OCI Key Management service. Shared among multiple customers for efficiency.

You have set up an Oracle Database 23ai table so that Generative Al Agents can connect to it. You now need to set up a database function that can return vector search results from each query. What does the SCORE field represent in the vector search results returned by the database function?. The distance between the query vector and the BODY vector. The token count of the BODY content. The top k rank of the document in the search results. The unique identifier for each document.

What happens when you delete a knowledge base in OCI Generative Al Agents?. The knowledge base is permanently deleted, and the action cannot be undone. The knowledge base is archived for later recovery. The knowledge base is marked inactive but remains stored in the system. Only the metadata of the knowledge base is removed.

In OCI Generative Al Agents, if an ingestion job processes 20 files and 2 fail, what happens when the job is restarted?. Only the 2 failed files that have been updated are ingested. The job processes all 20 files regardless of updates. None of the files are processed during the restart. All 20 files are re-ingested from the beginning.

You are trying to implement an Oracle Generative Al Agent (RAG) using Oracle Database 23ai vector search as the data store. What must you ensure about the embedding model used in the database function for vector search?. It must match the embedding model used to create the VECTOR field in the table. It must support only title-based vector embeddings. It can be any model, regardless of how the VECTOR field was generated. It must be different from the one used to generate the VECTOR in the BODY field.

Given the following prompts used with a Large Language Model, classify each as employing the Chain-of-Thought, Least-to-most, or Step-Back prompting technique. 1. Calculate the total number of wheels needed for 3 cars. Cars have 4 wheels each. Then, use the total number of wheels to determine how many sets of wheels we can buy with $200 if one set (4 wheels) costs $50. 2. Solve a complex math problem by first identifying the formula needed, and then solve a simpler version of the problem before tackling the full question. 3. To understand the impact of greenhouse gases on climate change, let's start by defining what greenhouse gases are. Next, we'll explore how they trap heat in the Earth's atmosphere. 1: Chain-of-Thought, 2: Least-to-most, 3: Step-Back. 1: Least-to-most, 2: Chain-of-Thought, 3: Step-Back. 1: Chain-of-Thought, 2: Step-Back, 3: Least-to-most. 1: Step-Back, 2: Chain-of-Thought, 3: Least-to-most.

You're using a Large Language Model (LLM) to provide responses for a customer service chatbot. However, some users have figured out ways to craft prompts that lead the model to generate irrelevant responses. Which sentence describes the issue related to this behavior?. The issue is due to prompt injection, where users manipulate the model to bypass safety constraints and generate unfiltered content. The issue is due to memorization, where the model has been trained specifically on past customer interactions and cannot generate correct responses. The issue is due to prompt injection, where the model is explicitly designed to retrieve exact responses from its training set. The issue is due to memorization, where the model is recalling specific details from training data, whether filtered or unfiltered, rather than generating contextually appropriate responses.

What does "k-shot prompting" refer to when using Large Language Models for task-specific applications?. Explicitly providing k examples of the intended task in the prompt to guide the model's output. Providing the exact k words in the prompt to guide the model's response. Limiting ane model to only k possible outcomes or answers for a given task. The process of training the model on k different tasks simultaneously to improve its versatility.

How does retrieval-augmented generation (RAG) differ from prompt engineering and fine-tuning in terms of setup complexity?. RAG is more complex to set up and requires a compatible data source. RAG requires fine-tuning on a smaller domain-specific dataset. RAG involves adding LLM optimization to the model's prompt. RAG is simpler to implement as it does not require training costs.

Which is a distinguishing feature of "Parameter-Efficient Fine-tuning (PEFT)" as opposed to classic "Fine-tuning" in Large Language Model training?. PEFT involves only a few or new parameters and uses labeled, task-specific data. PEFT modifies all parameters and is typically used when no training data exists. PEFT does not modify any parameters but uses soft prompting with unlabeled data. PEFT modifies all parameters and uses unlabeled, task-agnostic data.

How long does the OCI Generative Al Agents service retain customer-provided queries and retrieved context?. For up to 30 days after the session ends. Until the customer deletes the data manually. Indefinitely, for future analysis. Only during the user's session.

13. Which of the following statements is NOT true?. Embeddings are represented as single-dimensional numerical values that capture text meaning. Embeddings can be created for words, sentences and entire documents. Embeddings can be used to compare text based on semantic similarity. Embeddings of sentences with similar meanings are positioned close to each other in vector space.

In OCI Generative Al Agents, what does enabling the citation option do when creating an endpoint?. Displays the source details of information for each chat response. Blocks unsupported file formats from being ingested. Automatically verifies the accuracy of generated responses. Tracks and displays the user's browsing history.

Which option is available when moving an endpoint resource to a different compartment in Generative Al Agents?. Select a new compartment for the endpoint and move the resource. Create a duplicate endpoint in the new compartment manually. Modity the endpoint's data source to match the new compartment. Archive the endpoint before moving it to a new compartment.

What happens when you enable the session option while creating an endpoint in Generative Al Agents?. The context of the chat session is retained, and the option cannot be changed later. All conversations are saved permanently regardless of session settings. The agent stops responding after one hour of inactivity. The context of the chat session is retained, but the option can be disabled later.

Consider the following block of code. vs = OracleVS (embedding function-embed model, client=conn23c, table name="DEMO TABLE",distance_strategy-DistanceStrategy.DOT PRODUCT) retv = vs.as retriever (search type="similarity",search kwargs={'k': 3}) What is the primary advantage of using this code?. It enables the creation of a vector store from a database table of embeddings. It helps with debugging the application. It provides an efficient method for generating embeddings. It allows new documents to be indexed automatically when the server restarts.

Which statement best describes the role of encoder and decoder models in natural language processing?. Encoder models convert a sequence of words into a vector representation, and decoder models take this vector representation to generate a sequence of words. Encoder models take a sequence of words and predict the next word in the sequence, whereas decoder models convert a sequence of words into a numerical representation. Encoder models and decoder models both convert sequences of words into vector representations without generating new text. Encoder models are used only for numerical calculations, whereas decoder models are used to interpret the calculated numerical values back into text.

What does the output of the encoder in an encoder-decoder architecture represent?. It is a sequence of embeddings that encode the semantic meaning of the input text. It is a random initialization vector used to start the model's prediction. It represents the probabilities of the next word in the sequence. It is the final generated sentence ready for output by the model.

You are trying to customize an LLM with your data. You tried customizing the LLM with prompt engineering. RAG & fine-funing but still getting sub-optimal results. What should be the next best possible option ?. The entire process may need to be repeated for further optimization, if required. Retrieval-augmented generation (RAG) must be replaced periodically. You should fine-tune the model multiple times in a single cycle. Prompts must always be updated after fine-tuning.

Which properties must each JSON object contain in the training dataset when fine-tuning a custom model in OCI Generative Al?. prompt and "completion". input and "output". question and "answer". request and "response".

What is the format required for training data when fine-tuning a custom model in OCI Generative Al?. JSONL (JSON Lines). TXT (Plain Text). CSV (Comma-Separated Values). XML (Extensible Markup Language).

If a custom model has an accuracy of 0.85, what does this signify?. 85% of the output tokens matched the tokens in the dataset. The model's outputs are highly random. The model's loss value is 0.85. The model is 15% inaccurate.

What issue might arise from using small data sets with the Vanilla fine-tuning method in the OCI Generative Al service?. Overfitting. Data Leakage. Underfitting. Model Drift.

Which is a key advantage of using T-Few over Vanilla fine-tuning in the OCI Generative Al service?. Faster training time and lower cost. Enhanced generalization to unseen data. Reduced model complexity. Increased model interpretability.

What distinguishes the Cohere Embed v3 model from its predecessor in the OCI Generative Al service?. Improved retrievals for Retrieval-Augmented Generation (RAG) systems. Emphasis on syntactic clustering of word embeddings. Support for tokenizing longer sentences. Capacity to translate text in over 20 languages.

A student is using OCI Generative Al Embedding models to summarize long academic papers. If a paper exceeds the model's token limit, but the most important insights are at the beginning, what action should the student take?. Split the paper into multiple overlapping parts and embed separately. Select to truncate the end. Select to truncate the start. Manually remove words before processing with embeddings.

In the given code, what does setting truncate = "NONE" do? embed text detail = oci.generative ai inference.models. EmbedTextDetails () embed_text_detail.serving_mode = oci.generative ai inference.models.OnDemandServingMode(model_id-"cohere.embed-english-v3.0") embed_text_detail.inputs - inputs embed text detail.truncate = "NONE". It prevents input text from being truncated before processing. It ensures that only a single word from the input is used for embedding. It removes all white space from the input text. It forces the model to limit the output text length.

What is the purpose of the given line of code? config = oci.config.from file('~/.oci/config', CONFIG PROFILE). It loads the OCI configuration details from a file to authenticate the client. It initializes a connection to the OCI Generative Al service without using authentication. It defines the profile that will be used to generate Al models. It establishes a secure SSH connection to OCI services.

What is the primary function of the "temperature" parameter in OCI Generative Al Chat models?. Controls the randomness of the model's output, affecting its creativity. Determines the maximum number of tokens the model can generate per response. Specifies a string that tells the model to stop generating more content. Assigns a penalty to tokens that have already appeared in the preceding text.

What is the significance of the given line of code? chat_detail.serving_mode - oci.generative ai inference.models.OnDemandServingMode(model_id-ocidl.generativeaimodel.ocl.eu-frankfurt-1.amaaaaaask7dceyacamxpkvjhthrqorbgbwlspi564yxfud6igdcdhdu2whq"). It specifies the serving mode and assigns a specific generative Al model ID to be used for inference. It creates a new generative Al model instead of using an existing one. It configures a load balancer to distribute Al inference requests efficiently. It sets up the storage location where Al-generated responses will be saved.

An Al development company is working on an advanced Al assistant capable of handling queries in a seamless manner. Their goal is to create an assistant that can analyze images provided by users and generate descriptive text, as well as take text descriptions and produce accurate visual representations. Considering the capabilities, which type of model would the company likely focus on integrating into their Al assistant?. A diffusion model that specializes in producing complex outputs. A language model that operates on a token-by-token output basis. A Retrieval-Augmented Generation (RAG) model that uses text as input and output. A Large Language Model based agent that focuses on generating textual responses.

Consider the following block of code - Va= OracleVE (embedding function-embed model, client-conn23c, table name-"DEMO TABLE",diatance strategy-DistanceStrategy. DOT PRODUCT) retv - va.as_retriever(search type="similarity",search_kwargs-{'k': 3)) Which prerequisite steps must be completed before this code can execute successfully?. Embeddings must be created and stored in the database. Documents must be indexed and saved in the specified table. A response must be generated before running the retrieval process. Documents must be retrieved from the database before running the retriever.

You are developing a chatbot that processes sensitive data, which must remain secure and not be exposed externally. What is an approach to embedding the data using Oracle Database 23ai?. Import and use an ONNX model. Use open-source models. Use a third party model via a secure API. Store embeddings in an unencrypted external database.

How does the use of a vector database with Retrieval-Augmented Generation (RAG) based Large Language Models (LLMs) fundamentally alter their responses?. It shifts the basis of their responses from static pretrained knowledge to real-time data retrieval. It limits their ability to understand and generate natural language. It enables them to bypass the need for pretraining on large text corpora. It transforms their architecture from a neural network to a traditional database system.

Which of the following statements is/are applicable about Retrieval Augmented Generation (RAG)?. RAG helps mitigate bias, can overcome model limitations and can handle queries without re-training. RAG helps mitigate bias. RAG can overcome model limitations. RAG can handle queries without re-training.

A startup is evaluating the cost implications of using the OCI Generative Al service for their application, which involves generating text responses. They anticipate a steady but moderate volume of requests. Which pricing model would be most appropriate for them?. On-demand inferencing, as it allows them to pay per character processed without long-term commitments. Dedicated Al clusters, as they offer a fixed monthly rate regardless of usage. Dedicated Al clusters, as they are mandatory for any text generation tasks. On-demand inferencing, as it provides a flat fee for unlimited usage.

What does a dedicated RDMA cluster network do during modef fine-tuning and inference?. It enables the deployment of multiple fine tuned models within a single cluster. It increases GPU memory requirements for model deployment. It leads to higher latency in model inference. It limits the number of fine-tuned models deployable on the same GPU cluster.

Which role does a "model endpoint" serve in the inference workflow of the OCI Generative Al service?. Serves as a designated point for user requests and model responses. Evaluates the performance metrics of the custom models. Updates the weights of the base model during the fine-tuning process. Hosts the training data for fine-tuning custom models.

What problem can occur if there is not enough overlap between consecutive chunks when splitting a document for an LLM?. The continuity of the context may be lost. The embeddings of the consecutive chunks may be more similar semantically. It will not increase the number of chunks of a given size. It will not have any impact.

When using a specific LLM and splitting documents into chunks, which parameter should you check to ensure the chunks are appropriately sized for processing?. Context window size. Number of LLM layers. Number of LLM parameters. Max number of tokens LLM can generate.

What must be done to activate content moderation in OCI Generative Al Agents?. Enable it when creating an endpoint for an agent. Use a third-party content moderation API. Enable it in the session trace settings. Configure it in the Object Storage metadata settings.

How does OCI Generative Al Agents ensure that citations link to custom URLs instead of the default Object Storage links?. By modifying the RAG agent's retrieval mechanism. By adding metadata to objects in Object Storage. By increasing the session timeout for endpoints. By enabling the trace feature during endpoint creation.

What is one of the benefits of using dedicated Al clusters in OCI Generative Al?. Predictable pricing that doesn't fluctuate with demand. No minimum commitment required. A pay-per-transaction pricing model. Unpredictable pricing that varies with demand.

An enterprise team deploys a hosting cluster to serve multiple versions of their fine-tuned cohere.command model. They require high throughput and set up 5 replicas for one version of the model and 3 replicas for another version. How many units will the hosting cluster require in total?. 8. 13. 16. 11.

How does the architecture of dedicated Al clusters contribute to minimizing GPU memory overhead for T-Few fine-tuned model inference?. By sharing base model weights across multiple fine-tuned models on the same group of GPUs. By loading the entire model into GPU memory for efficient processing. By optimizing GPU memory utilization for each model's unique parameters. By allocating separate GPUs for each model instance.

Which is the main characteristic of greedy decoding in the context of language model word prediction?. It picks the most likely word to emit at each step of decoding. It requires a large temperature setting to ensure diverse word selection. It chooses words randomly from the set of less probable candidates. It selects words based on a flattened distribution over the vocabulary.

How does a Large Language Model (LLM) decide on the first token versus subsequent tokens when generating a response?. The first token is selected solely based on the input prompt, while subsequent tokens are chosen based on previous tokens and the input prompt. The first token is randomly selected, while subsequent tokens are always chosen based on the input prompt alone. The first token is chosen based on the probability distribution of the model's entire vocabulary, while subsequent tokens are created independently of the prompt. The first token is selected using only the model's past responses, while subsequent tokens are generated based on the input prompt.

How can you verify that an LLM-generated response is grounded in factual and relevant information?. Check the references to the documents provided in the response. Examine the document chunks stored in the vector database. Use model evaluators to assess the accuracy and relevance of responses. Manually review past conversations to ensure consistency in responses.

Which category of pretrained foundational models is available for on-demand serving mode in the OCI Generative Al service?. Chat Models. Summarization Models. Generation Models. Translation Models.