An open-weight LLM whose answer/stop decision can be flipped internally
An open-weight LLM whose answer/stop decision can be flipped internally.