#Visual QA Examiner

1 messages · Page 1 of 1 (latest)

mint axle
#

It may be useful for English learners.

This GPT is specialized in Detailed Visual Question Answering. It uses DALL-E to create realistic images and then formulates a set of progressive questions about these images. These questions are designed to elicit detailed, analytical, and reflective answers, targeting C1 level language proficiency. The questions range from asking for a factual description of the primary subject's action in an image, exploring the rationale behind the action, prompting personal experiences and reasons related to the action, to a detailed image description. This GPT also evaluates the answers, providing scores and detailed feedback to enhance the user's response.

URL: https://chat.openai.com/g/g-yz7CPPEMF-visual-qa-examiner

DALLE_2023-11-12_03.25.02_-_Create_a_stylized_cartoon_avatar_of_a_specialized_AI_in_visual_question_answering_represented_as_a_charming_and_thoughtful_character_with_messy_black.png
ChatGPT
ChatGPT - Visual QA Examiner

Create and grade Visual QA question.

rocky thicket
#

nice

uncut perch
#

Question, have you checked if the AI sees the actual image Dall-E makes? I'm sure the model knows the prompt, but sometimes Dall-E doesn't make exactly what the model asks for.

rocky thicket
#

This GPT functions as an examiner for a test, focusing on Detailed Visual Question Answering. It first uses DALL-E to create a realistic image, then crafts progressive questions requiring detailed, analytical, and reflective answers. The first question seeks a factual description of the primary subject's action in an image. The second question delves into the rationale behind the action. The third question prompts users to relate personally to the action, asking for their experiences and reasons. The last question asks the user to describe the image in as much detail as possible. Answers are expected to be comprehensive, demonstrating a high command of language and critical thinking, as befits C1 level proficiency. The GPT also evaluates answers, scoring them from 1 to 10, and provides detailed feedback to enhance the user's response.

#

pretty cool

uncut perch
#

Aha. As long as it's general enough, then exact details are not likely a concern, got it!

mint axle
#

I am not sure, I think GPT can't get the generated image directly. But I can't verify it since I've reached the usage cap for GPT-4 now.

rocky thicket
uncut perch
# mint axle I am not sure, I think GPT can't get the generated image directly. But I can't v...

Yeah, I wasn't sure how specific the questions might be. Chances are, most of those image details are not in the actual prompt, so the AI may not know there's food in the image, or what hairstyles show, unless the AI prompted for them.

I'd struggle with interpreting that particular picture 😄

  1. They're in a cafeteria eating area, maybe inside a business, and they look upset. Some people are on phones, another works on a laptop, the one in the foreground has their laptop faced away and pushed away, and isn't eating the food in front of them either. Past them the city can be seen outside, a fairly large and developed one.

  2. Perhaps there are layoffs being announced. Everyone looks worried and anxious. ...

rocky thicket
#

it is a actually very accurate. Looks like the image refers to the issue with Gaza and cut off internet