C1000-185試験無料問題集「IBM watsonx Generative AI Engineer - Associate 認定」

A. Fine-tune the prompt template to emphasize grammar

B. Lower the learning rate during the tuning phase

C. Use a higher temperature during the generation process

D. Increase the model's training dataset size

A. Boolean retriever, which uses logical operators (AND, OR, NOT) to filter documents based on keyword presence.

B. Hierarchical retriever, which first applies keyword-based filters and then ranks documents using machine learning models.

C. Dense retriever, which converts both the query and documents into vectors and retrieves based on similarity in the vector space.

D. Exact match retriever, which retrieves documents that exactly match the query string.

A. Embeddings represent the search space for the retriever model, allowing the system to retrieve semantically relevant information based on input queries.

B. Embeddings are used to directly generate the textual responses in the output.

C. Embeddings are pre-trained generative models that augment the retrieval step by generating new query variations.

D. Embeddings are only used in fine-tuning generative models and play no role in the retrieval process.

A. Use only watsonx.ai's built-in APIs and SDKs for integration, as LangChain is not required for chaining multiple prompts.

B. Directly integrate the external APIs with watsonx.ai without any intermediate framework, since LangChain would add unnecessary overhead.

C. Implement LangChain to handle complex multi-step workflows, using watsonx.ai's APIs to generate responses at specific stages in the chain.

D. Write custom scripts to manage all prompt sequences manually, leveraging watsonx.ai's SDK to call the generative AI model at every step.

A. Embeddings store the entire content of documents, which is then directly passed to the generative model.

B. Embeddings reduce the size of the generative model by compressing the parameters into a smaller representation.

C. Embeddings ensure that only syntactically correct documents are retrieved, without regard to semantic content.

D. Embeddings are used to retrieve relevant documents by calculating the semantic similarity between user queries and the stored documents.

A. Use the smallest possible model to reduce computational costs, regardless of the use case.

B. Use a large language model fine-tuned for email response generation for both tasks, as the fine-tuning process will enable the model to handle creative tasks as well.

C. Choose a domain-specific model for automating email responses and a more generalized model with creative capabilities for generating marketing copy.

D. A single model should be used for both use cases since Watsonx can handle any type of text generation task equally well.

A. The model exhibits acceptable performance but occasionally generates off-topic responses unrelated to financial data.

B. The model's accuracy fluctuates based on the length of the input text.

C. The model generates output that is highly relevant to general topics but often misinterprets industry-specific terms like "leverage" or "derivative."

D. The model produces a high number of tokens in each output, which increases the cost of usage.

A. The model might produce content that doesn't align with the cultural preferences of the user.

B. The model can generate outputs that are too general, failing to meet the specific needs of the user.

C. The model can unintentionally memorize and regurgitate personal information from the training data, leading to privacy violations.

D. The model can generate overly creative or non-factual responses, leading to brand reputation damage.

A. "Guide the user through scheduling a meeting by asking about their preferred date, time, and attendees."

B. "Provide a list of all upcoming events in the user's calendar and ask if they want to add another."

C. "Please provide the user's calendar availability."

D. "Ask the user when they want to schedule their meeting."

A. Set the stop sequence to "End of Response"

B. Set the stop sequence to "<|stop|>"

C. Set the stop sequence to "STOP"

D. Set the stop sequence to "\ n\ n"

A. Manually categorize each interaction and organize them into a taxonomy tree structure.

B. Tokenize the dataset before curating it and mapping it to the taxonomy tree.

C. Perform a spelling correction on the entire dataset to remove any language inconsistencies.

D. Convert all text to lowercase to ensure uniformity in the dataset.

A. "Troubleshoot the following issue: [Issue Description]. Offer step-by-step commands and reasons for failure."

B. "Give troubleshooting instructions for [Issue Description], including steps, commands, and reasons for failure."

C. "Provide step-by-step instructions for troubleshooting the issue: [Issue Description]. Include commands and reasons for failure."

D. "Provide detailed troubleshooting instructions for the issue: [Issue Description], with steps, commands, and potential reasons for failure."

A. Tuning Studio allows selective parameter updates, reducing the need to retrain the entire model for each new task.

B. Tuning Studio integrates with existing AI infrastructure to streamline model fine-tuning without requiring complex deployment processes.

C. Tuning Studio offers fine-tuning with minimal data, allowing users to adapt models to niche tasks without needing large datasets.

D. Tuning Studio enables real-time, dynamic adjustments to the model's architecture during inference to handle new tasks.

A. Embed privacy-sensitive heuristics in the model's prompt to guide its behavior

B. Leverage regular expressions to filter out personal information in the prompt

C. Train the model exclusively on datasets that contain no personal information

D. Use a post-generation filter to remove any text that appears to be personal information

A. Train the model to generate creative financial advice tailored to each customer.

B. Focus on improving the model's ability to generate financial market forecasts.

C. Optimize for natural language generation to improve customer engagement.

D. Fine-tune the model for information retrieval tasks related to customer accounts.