5 ms·
Most of the blame that LLMs get is because they are used wrong. LLM is a good information transformation engine. But if I supply it with messy or wrong informa
by abdullin 2y ago
Most of the blame that LLMs get is because they are used wrong.
LLM is a good information transformation engine. But if I supply it with messy or wrong information in the context, then I’ll end up with hallucinations.
This is why most of the AI systems tend to produce junk. LLM part is good, but the underlying system just provides it with noisy and meaningless data.