Lupin
@lp1.bsky.social
💻 Hacker with a short attention span // 📚 Information Security Trainer // 🎓 Epitech Alumi // http://keybase.io/lp1 // Pentest+ / CySA+
I'm quite sure even today's state of the art LLMs cannot do either. Best case scenario, they would actually generate an output with a valid way of counting if you gave a precise enough input, because that would be the stastically most probable output, but they are far from doing any kind of thinking
November 6, 2024 at 2:34 PM
I'm quite sure even today's state of the art LLMs cannot do either. Best case scenario, they would actually generate an output with a valid way of counting if you gave a precise enough input, because that would be the stastically most probable output, but they are far from doing any kind of thinking
My first test, I guess I wasn't expecting more from a LLM. A reminder that they are language models built to generate probable outputs, not to reason, deepseek is no exception.
November 6, 2024 at 1:45 PM
My first test, I guess I wasn't expecting more from a LLM. A reminder that they are language models built to generate probable outputs, not to reason, deepseek is no exception.