It feels like it's sort of it's own thing. LLMs are really good at morphing or fuzzy finding.
An interesting example – I had a project where I needed to parse out addresses and dates in a document. However, the address and date formats were not standardized across documents. Utilizing LLMs was way easier then trying to regex or pattern match across the text.
But if you're trying to take a text document and break it down into some sort of a structured output, the outcome using LLMs will be much more variable.
Comments
It feels like it's sort of it's own thing. LLMs are really good at morphing or fuzzy finding.
An interesting example – I had a project where I needed to parse out addresses and dates in a document. However, the address and date formats were not standardized across documents. Utilizing LLMs was way easier then trying to regex or pattern match across the text.
But if you're trying to take a text document and break it down into some sort of a structured output, the outcome using LLMs will be much more variable.