Billing for Oracle Cloud Infrastructure (OCI) Generative AI dedicated clusters is based on "unit hours." A dedicated AI cluster specifically for fine-tuning is composed of 2 units. The total consumption is calculated by multiplying the number of units in the cluster by the number of hours it is active.

In this scenario:

- Number of units for a fine-tuning cluster = 2

- Active duration = 10 hours

- Total unit hours = 2 units 10 hours = 20 unit hours.

A. 25 unit hours: This is an incorrect calculation. A fine-tuning cluster does not consist of 2.5 units.

B. 40 unit hours: This would be the correct calculation for a hosting dedicated AI cluster, which has 4 units (4 units 10 hours), not a fine-tuning cluster.

D. 30 unit hours: This is an incorrect calculation. A fine-tuning cluster does not consist of 3 units.

1. Oracle Cloud Infrastructure Documentation, "Generative AI - Dedicated AI Clusters": This document specifies the composition of different cluster types. It states, "A fine-tuning cluster has 2 units. A hosting cluster has 4 units." This directly supports the use of '2' as the multiplier for a fine-tuning cluster.

Reference Location: Oracle Cloud Infrastructure Documentation > AI and Machine Learning > Generative AI > Dedicated AI Clusters.

2. Oracle Cloud Infrastructure Pricing, "Generative AI": The official pricing page details the billing model. Under the "Dedicated AI Clusters" section, it explicitly states, "Fine-tuning: Billed per hour for each fine-tuning cluster. Each cluster has 2 units." This confirms the billing metric and the number of units.

Reference Location: Oracle Cloud website > Cloud > Cloud Pricing > Generative AI.

Prompt templates are structured, reusable text formats that serve as a blueprint or "recipe" for generating prompts. They contain placeholders (variables) that are filled in with specific data or context before being sent to a language model. This approach ensures consistency, guides the model's response structure, and simplifies the process of creating effective prompts for various tasks. By defining the context, instructions, and input data format, templates guide the model to produce more predictable and relevant outputs.

A. Prompt templates are text-based structures with placeholders, not complex algorithms that require compilation like source code.

C. The fundamental purpose of a template is to be modified and customized by filling in variables with specific context or data.

D. Prompt templates are designed primarily for textual content, as this is the native input format for language models.

1. Official Vendor Documentation: Oracle Cloud Infrastructure (OCI) Documentation, "Generative AI - Prompt Engineering". The documentation explains that a prompt consists of instructions and context, which can be structured for reuse. This structured format with placeholders is functionally a template or recipe. The guide states, "A good prompt gives the model enough context... You can also provide few-shot examples in the prompt to guide the model." This practice of creating a structured, reusable format is the essence of a template. (See OCI Documentation > AI and Machine Learning > Generative AI > Key Concepts > Prompt Engineering).

2. Academic Publication: Liu, P., Yuan, W., Fu, J., et al. (2023). "Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing." ACM Computing Surveys, 55(9), Article 195, p. 6. The paper discusses "Template Construction," describing templates as textual strings containing slots (placeholders) for input text. It states, "The first step of prompt-based learning is to design a template, which is a textual string that has two slots...". This definition aligns directly with the concept of a predefined recipe. DOI: https://doi.org/10.1145/3560815

3. University Courseware: Stanford University, CS324 - Large Language Models, Winter 2022, "Lecture 3: Prompting, Instruction Finetuning". The course materials discuss how prompts are structured to elicit desired behavior. The concept of creating a format (e.g., "Translate English to French: {englishtext} =>") is presented, which is a form of templating or a recipe for a specific task.

In the context of Large Language Models (LLMs), "hallucination" refers to the model generating content that is factually incorrect, nonsensical, or not grounded in the provided source data, yet presented with a high degree of confidence as if it were true. This phenomenon occurs because LLMs are probabilistic systems designed to predict the next most likely sequence of words based on their training data, rather than accessing a knowledge base to verify facts. As a result, they can invent details, fabricate sources, or create plausible-sounding but entirely false information. Mitigating hallucinations is a primary challenge in ensuring the reliability and trustworthiness of generative AI applications.

A. This describes creativity. While related, "hallucination" specifically refers to the undesirable generation of falsehoods, not desired imaginative output.

B. This is incorrect. Hallucination is a significant problem to be mitigated, not a technique used to improve model performance.

C. This describes a capability of multimodal models (e.g., image-to-text generation), not the definition of the term "hallucination."

1. Oracle Cloud Infrastructure (OCI) Documentation: In the official OCI Generative AI service FAQs, Oracle defines the term directly: "What is a hallucination? A hallucination is when a large language model (LLM) generates text that is nonsensical or not factually correct. Mitigating hallucinations is a key challenge in the field of generative AI."

Source: Oracle Cloud Infrastructure Documentation, "Generative AI - FAQs," Section: "What is a hallucination?".

2. Academic Publication: A comprehensive survey on the topic defines the phenomenon as follows: "Despite their remarkable capabilities, LLMs are prone to generating content that is nonsensical, unfaithful to the provided source, or otherwise untruthful. This phenomenon is commonly referred to as hallucination."

Source: Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., ... & Fung, P. (2023). Survey of Hallucination in Natural Language Generation. ACM Computing Surveys, 55(12), Article 357, p. 2. DOI: https://doi.org/10.1145/3571730

3. University Courseware: Stanford University's Human-Centered AI Institute (HAI) explains the concept in its foundational materials: "When a large language model (LLM) generates text that is nonsensical or untethered from reality, it is said to be hallucinating... The model is simply generating a statistically likely sequence of words, and sometimes that sequence doesn’t align with our reality."

Source: Stanford University Human-Centered AI (HAI), "How to Tame AI Hallucinations," Published May 16, 2023.

The frequency penalty is a parameter used to control the output of a language model during text generation. Its purpose is to decrease the likelihood of the model selecting tokens that have already appeared in the generated text or the prompt. The penalty is proportional to the frequency of the token's appearance. By applying a positive frequency penalty, the model is discouraged from repeating the same words or phrases, which helps to reduce redundancy and improve the overall diversity and readability of the output.

A. This is the opposite of a frequency penalty's function; it would encourage, not discourage, repetition.

C. A frequency penalty does not reward new tokens; it only penalizes existing ones. The concept of penalizing new tokens is related to the 'presence penalty'.

D. The penalty is deterministic, not random. It is applied systematically to tokens based on their exact frequency count in the preceding text.

1. Oracle Cloud Infrastructure Documentation, Generative AI Service API Reference. In the GenerateTextDetails schema, the frequencyPenalty parameter is defined as: "To reduce the likelihood of repeating tokens, assign a positive value to this parameter. The value is added to the log-probability of every token that has already appeared in the prompt or in the generated text." This directly confirms that the penalty is based on the number of times a token has been used. (See oracle-cloud-infrastructure-go-sdk/generativeai/GenerateTextDetails.go or the REST API documentation for the Generative AI service).

2. Oracle Cloud Infrastructure Documentation, Generative AI Service. The service documentation explains the use of Cohere models, whose parameters are mirrored in the OCI service. The Cohere documentation for the Generate endpoint describes frequencypenalty as: "The penalty to apply to tokens based on their frequency in the generated text. A higher value will discourage repetition." This aligns with penalizing tokens based on their usage count. (See Cohere API Documentation, Generate endpoint parameters).

A stop sequence is a user-defined string of characters that instructs the text generation model to halt its output. When a period (.) is specified as a stop sequence, the model will generate text token by token until it produces a period. Upon generating the period, the model immediately ceases generation, effectively completing the first sentence. This termination occurs irrespective of the maximum token limit, provided the period is generated before that limit is reached.

A. This is incorrect because the explicit purpose of a stop sequence is to be recognized and acted upon, not ignored.

B. This is the opposite of the intended behavior. A stop sequence halts generation, it does not prompt the model to continue.

C. A period marks the end of a sentence, not necessarily a paragraph. The model stops at the first instance of the sequence, not at a larger text structure.

1. Oracle Cloud Infrastructure (OCI) Documentation - Generative AI: The GenerateText API operation includes the stopSequences parameter. The documentation states: "A list of strings that specifies where the model should stop generating tokens. If the model generates a string that matches one of the strings in the list, it stops generating tokens." This confirms that the model's generation is terminated upon encountering the specified sequence.

Source: Oracle Cloud Infrastructure Documentation, REST API for Generative AI Inference, GenerateText operation, GenerateTextDetails object, stopSequences parameter description. (Accessed via Oracle's official documentation portal).

2. Stanford University - CS224N: NLP with Deep Learning: Course materials on language model decoding strategies explain that generation can be controlled by various parameters, including termination conditions. A user-defined stop sequence is a common termination condition that overrides the maximum length/token limit if encountered first.

Source: Stanford University, CS224N: Natural Language Processing with Deep Learning, Winter 2023, Lecture 8: "Language Generation", section on Decoding Algorithms and controlling output. (Conceptual reference).