Expected Parrot
Using Images
expectedparrot.com
Using Images
These slides show how to create scenarios for images that you want to use as variables with your questions. The same steps can be used for other types of content as well (text, videos, etc.)
Step by step:
Everything you create is reproducible in code, shareable, and exportable.
Open the scenarios modal
Select the dropdown option to upload images
Format the scenarios based on how many images you want to use with a question at once. A scenario can contain a single image or multiple images where each image is stored in a separate field.
Give the fields unique names for reference when adding them to questions
Name your scenarios and save them (they are automatically added to your Content for future use)
Create questions
Add scenario fields to your question texts
Optionally add agent personas to answer the questions (you can also just have models answer directly)
Agent personas can consist of any information you want the model to have in answering the questions
Select the vision model(s) that you want to use to answer the questions
You can use one or multiple models at once to compare performance
Run the survey
Inspect, export, and share results
Select the option to clone the survey to iterate on it or generate a human survey
Select the option to “Run with Humans” to generate a web-based survey
Share the survey link or QR code with respondents
Responses appear automatically