can you explain a bit more about the issue? It sounds like the AI is ignoring the max token output limit? are you using koboldcpp or lmstudio?
