6 ms·
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.
by rck 2y ago
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.
- Havoc 2y agoThat’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