Uncensored LLMs in agent loops: 31 models, 5 tasks, most fail half the time (openuncensored.com)

[flagged]

Without posting how you implemented the task loop and the parameters and model instructions or behaviours, this is a waste of bytes.