Google

When prompting a Large Language Model for a novel or unfamiliar classification task, providing a detailed description of the problem is crucial for achieving accurate results. This description should include specific details such as the classification standards to be used, ensuring the model fully understands the task's requirements.

Using Descriptive Prompts for Novel Classification Tasks

Few-shot learning is a technique for adapting a large language model to new tasks by providing a small number of demonstrations in the prompt. These demonstrations establish a pattern of input-to-output mappings, which the model then attempts to follow when making predictions on new inputs. This approach contrasts with zero-shot learning, where no examples are given, and traditional training, which requires a large dataset.

Few-Shot Learning

A software development team is using a large language model to sort user feedback for their new video editing application. They want to classify each piece of feedback into one of three custom categories: 'UI/UX Suggestion', 'Bug Report', or 'Feature Request'.

The team uses the following prompt: `Classify the user feedback below.`

They find that the model's classifications are inconsistent and often incorrect, frequently miscategorizing bug reports as feature requests. Analyze why this prompt is ineffective for this specific task. Then, describe two distinct, specific improvements that could be made to the prompt's description to help the model perform the classification more accurately, without providing examples of classified feedback.

Improving a Prompt for a Novel Classification Task

A software development team wants to use a large language model to automatically categorize user feedback for their new, specialized 3D architectural modeling software. They need to sort the feedback into three specific, internally-defined categories: 'Bug Report', 'Feature Request', and 'Usability Issue'. They have drafted the following prompt to give to the model for each piece of feedback:

`'Classify the following user feedback into one of these categories: Bug Report, Feature Request, or Usability Issue.'`

Evaluate the effectiveness of this prompt for this specific task. Explain why it is likely to produce inaccurate or inconsistent results and what key information is missing.

Evaluating a Prompt for a Custom Classification Task

A data scientist wants to use a large language model to categorize internal company documents into three newly-defined, specific categories: 'Alpha Project Brief', 'Beta Project Brief', and 'Gamma Project Brief'. The model has not been specifically trained on this internal classification system. Which of the following prompts is best designed to achieve the most accurate and consistent results for this task?

Learn Before

Related