alt Hacker NewsIt's interesting that, at least through ollama, it appears impossible to get DeepSeek-R1 to give concise answers to questions that don't require any reasoning.
It's interesting that, at least through ollama, it appears impossible to get DeepSeek-R1 to give concise answers to questions that don't require any reasoning.
That’s the nature of LLMs. They can’t really think ahead to „know“ whether reasoning is required. So if it’s tuned to spit out reasoning first then that’s what it’ll do