Jinyibruceli3 days ago
I'm curious how you're handling the self-evolving part in practice. When I built browser automation agents with Claude API last year, the hardest problem wasn't the initial task execution - it was knowing when to stop iterating vs when the agent had legitimately improved its approach. Does autodidact have any safeguards against the agent just spinning in circles rewriting its own prompts, or do you rely on the user to define success criteria upfront?
- waterbuffaloai2 days ago
  Yes, this is the exact engineering challenge I have been trying to solve. Currently, the improvements only include memories and skills update, for v1.5, this is what I am trying to solve too. Could you try the current version and let me know what you think?
hoansdz2 days ago
What happens if the model continuously hallucinates or provides wrong answers until the user give up ? Will the AI trust those wrong answers and save it ?
- waterbuffaloai2 days ago
  users have option to type /wrong, then the local model will ask cloud model for the answer and learn from it. Could you try the current version and let me know your feedbacks?
aayushkumar1213 days ago
[dead]
waterbuffaloai3 days ago
[flagged]