C1000-185 IBM watsonx Generative AI Engineer v1 - Associate Practice Questions
Prepare for C1000-185 with more than an answer.
- Exam fee
- $200 USD
- Level
- Associate
- Valid for
- Lifetime (no expiration)
Domains covered on the exam 6
- Analyze and Design a Generative AI Solution15%
- Prompt Engineering16%
- Fine-Tuning31%
- Retrieval-Augmented Generation (RAG)17%
- Deployment13%
- Integration with Model Orchestration8%
- 1
An organization wants to deploy a fine-tuned model to a production environment in watsonx.ai. They require the endpoint to handle high concurrency with low latency. Which deployment type should be selected in the Deployment Space?
Show answer details
Correct answer: B
Online deployments create a REST API endpoint designed for real-time (low latency) inference requests. This is the correct choice for applications requiring immediate responses to user interactions. Batch deployment is for processing large volumes of data asynchronously.
- 2
You are integrating a Python application with watsonx.ai using the
ibm-watsonx-aiSDK. You need to authenticate to the service. Which two pieces of information are typically required to initialize theAPIClient? (Select TWO)Show answer details
Correct answer: A, B
To initialize the
APIClientin theibm-watsonx-aiSDK, you typically need an IBM Cloud API Key for authentication and the Project ID (or Space ID) to define the context where resources will be accessed or created. The URL is often required but the API Key and Project ID are the primary credentials/identifiers.To initialize the
APIClientin theibm-watsonx-aiSDK, you typically need an IBM Cloud API Key for authentication and the Project ID (or Space ID) to define the context where resources will be accessed or created. The URL is often required but the API Key and Project ID are the primary credentials/identifiers. - 3
In the context of RAG (Retrieval-Augmented Generation), what is the purpose of the 'Chunking' process during the data ingestion phase?
Show answer details
Correct answer: B
LLMs have limited context windows (e.g., 4k or 8k tokens). Chunking breaks large documents into smaller pieces so that: 1) The most relevant specific sections can be retrieved, and 2) The retrieved text fits into the prompt sent to the LLM. It is not primarily for compression or translation.
- 4
A team is building a complex AI agent using LangChain and watsonx.ai. The agent needs to perform multi-step reasoning where the output of one step determines the tool used in the next. Which prompting strategy is most suitable for this 'Reasoning and Acting' capability?
Show answer details
Correct answer: A
ReAct (Reason + Act) is a prompting paradigm where the model generates a thought (Reasoning) about what to do, then performs an action (Act), and observes the result. This is the standard pattern for autonomous agents that need to use tools and chain logic together.
- 5
True or False: In watsonx.ai, 'Soft Prompts' created during Prompt Tuning are human-readable text strings that are added to the beginning of the prompt.
Show answer details
Correct answer: B
Soft Prompts are NOT human-readable text. They are learnable vector embeddings (tensors of numbers) that are optimized during the tuning process. 'Hard Prompts' are the human-readable text instructions.
- 6
You are using
ilab(InstructLab) to generate synthetic data for model alignment. What is the command used to generate this data based on the taxonomy you have defined?Show answer details
Correct answer: A
The command
ilab data generateis used in the InstructLab workflow to generate synthetic question-answer pairs based on the skills and knowledge defined in the local taxonomy. This data is subsequently used for training. - 7
A healthcare provider wants to use generative AI to summarize patient notes. They have strict governance requirements regarding Hate, Abuse, and Profanity (HAP). Which feature of watsonx.ai should be enabled to automatically filter out unsafe content during generation?
Show answer details
Correct answer: A
AI Guardrails (specifically HAP detection) in watsonx.ai are designed to monitor inputs and outputs for Hate, Abuse, and Profanity. Enabling this feature ensures that the model declines to generate or filters out content that violates these safety policies.
- 8
You are configuring a 'Chain-of-Thought' prompt to help a model solve complex logic puzzles. Which structure best represents this technique?
Show answer details
Correct answer: B
Chain-of-Thought (CoT) prompting involves providing examples (few-shot) where the model is shown how to break down the problem into intermediate reasoning steps before arriving at the final answer. This significantly improves performance on complex reasoning tasks.
- 9
What is the primary advantage of using LoRA (Low-Rank Adaptation) over full fine-tuning when customizing a 70B parameter model?
Show answer details
Correct answer: B
LoRA freezes the pre-trained model weights and injects trainable rank decomposition matrices into each layer of the Transformer architecture. This means only a tiny fraction of parameters are trained, drastically reducing GPU memory usage and storage requirements compared to full fine-tuning.
- 10
An engineer is defining a new 'skill' in the InstructLab taxonomy. Which file type must they create to define the seed examples (questions and answers) for this skill?
Show answer details
Correct answer: A
In the InstructLab taxonomy structure, a
qna.yamlfile is used to define the seed examples for a new skill or knowledge. This YAML file contains the 'question' and 'answer' pairs (and optionally 'context' for knowledge) that the teacher model uses to generate synthetic data. - 11
A financial services firm is using InstructLab to customize an IBM Granite model for internal policy compliance checks. The goal is to teach the model two things: first, to recognize and classify new, firm-specific financial product names, and second, to follow a specific three-step reasoning process for evaluating compliance. How should the engineering team structure their taxonomy files in InstructLab to achieve this?
Show answer details
Correct answer: C
This is the correct approach. InstructLab distinctly separates 'knowledge' from 'skills'. Knowledge pertains to factual information ('what'), such as product names, definitions, and facts. Skills pertain to abilities and processes ('how'), such as following instructions, performing a reasoning process, or writing in a specific style. Therefore, the new financial product names belong in the knowledge taxonomy, while the structured reasoning process belongs in the skills taxonomy.
- 12
An engineer is implementing Low-Rank Adaptation (LoRA) to fine-tune a foundation model on a domain-specific dataset. To ensure training is both effective and resource-efficient, which TWO of the following are the most critical LoRA-specific hyperparameters to configure? (Select TWO)
Show answer details
Correct answer: B, D
The rank (r) determines the dimensionality of the trainable update matrices. It is the most critical LoRA hyperparameter, directly controlling the trade-off between model expressiveness (higher r) and the number of trainable parameters (lower r). A well-chosen rank is key to successful and efficient tuning.
Alpha (α) is a scaling factor for the LoRA updates. It is used in conjunction with the rank (r) to control the magnitude of the adaptation. The ratio of α/r is often kept constant, making α a crucial parameter to tune alongside r for optimal performance.
