Attempt #2: Few-Shot
We used a process that we’re calling Few-Shot. We gave a description of the image, some context of our goal, and the three recent winning captions to a computer model to generate the results.Back to Project
Our submission: “I want to try living outside of my comfort zone.”
Check out all three generated captions.
“I want to try living outside of my comfort zone.”
29% of readers thought it was funny.
“We need to start somewhere.”
12% of readers thought it was funny.
“That’s what we’re supposed to be wearing?”
13% of readers thought it was funny.
About this approach
To produce the captions this week we “prompted” the GPT-3 model, davinci-instruct-beta, through the OpenAI API Playground. It took around 20 attempts to get these three results.
Whereas last week we expected the model to fully draw from its general knowledge bank to understand what we meant by “funny” and “New Yorker caption” and even how a description might relate to a caption, this week we give the model a few examples to nudge it in the right direction.
Here is the full prompt:
Translate the given cartoon description into a funny caption for a New Yorker cartoon:
Cartoon description: People are walking on the sidewalk. Everyone is inside their own bubble.
Cartoon caption: "I want one that snows."
Cartoon description: There is a magician pulling a rabbit out of a hat on his balcony. He has an audience of people watching from their apartment windows.
Cartoon caption: "See? And you said you couldn't work from home."
Cartoon description: Two firemen are deciding which pole to go down. One has a circular opening, one has a square opening.
Cartoon caption: "Since when did the pizza delivery guy get his own pole?"
Cartoon description: People are walking on a busy sidewalk in front of apartments. Everyone is wearing a mask. An unmasked person is working at their desk, in the middle of the sidewalk, which is blocking the people walking.