A user provides the keywords 'haunted mansion, brave detective, hidden treasure' to a text generation model. The model produces a single sentence: 'A brave detective found treasure in a haunted mansion.' Analyze why this output, while factually incorporating the keywords, fails to demonstrate the key capability expected from a model designed for generating a lengthy, coherent narrative from a concise input.

Google

An illustrative example of long text generation is the task of creating a lengthy story based on a few keywords. In this case, the model takes a concise input (the keywords) and produces an extensive, coherent narrative as the output.

Example of Long Text Generation: Story Generation

A developer is testing a new text generation model's ability to create extensive, coherent narratives from minimal starting points. Which of the following test cases best demonstrates this specific capability?

Evaluating Story Generation Output

Based on the model's capabilities described in the case study, propose a concise input (prompt) that would be effective for generating the desired backstory. Justify why your proposed prompt is well-suited for this task.

Learn Before

Related