edsl
edsl copied to clipboard
Design, conduct and analyze results of AI-powered surveys and experiments. Simulate social science and market research with large numbers of AI agents and LLMs.
A few notes: 1. If there a validation error, we don't add the answer to the current_answers dictionary. This causes piped answers to fail. I think we probably do not...
1. it should have the same .by(models) functionality as the same as everything else - otherwise confusing 2. if i limit the number of turns to say, 6, there should...
Currently, we only support a single image sent to the LanguageModel via an base64 encoded image. However: 1. This is not how Gemini works 2. Gemini can support multiple formats...
Progress bar can be hard to see -- missing it when I scroll down, and it repeatedly dis/appears: https://github.com/user-attachments/assets/eb3f992e-f6be-4ffd-8ad3-1895dd2871c2
This would allow us combine human and AI agent responses into the same dataset.