About Playground¶
GALE's prompt Playground allows you to experiment and refine prompts to get the best performance from LLMs. You can test different prompts across various models - external, fine-tuned, or open-source. The Playground helps you identify each prompt's ideal model and configurations through an iterative process. This streamlined workspace enables you to optimize prompts rapidly for maximum model effectiveness.
Key Features¶
-
Create a Prompt Experiment: Create experiments by selecting a template or writing a custom prompt and adding examples. Link a dataset, add one or more LLMs to generate responses, and compare the results. Learn more.
-
Simple vs. Advanced Mode: Simple mode allows quickly generating output by adding a prompt and model. Advanced mode provides more control, enabling the addition of a prompt, system prompt, and examples, linking a dataset, and adding multiple models to generate and compare outputs. Learn more.
-
Reuse the Prompt Experiment: Copy the prompt as a Gen AI node directly in the Agent Flow builder or copy it for use elsewhere. You can also export/download the prompt and related data in CSV format. Learn more.
-
Restore a Prompt Experiment from History: Review the timeline of prompts and their corresponding generated outputs, and restore a prompt version if needed. Learn more.