Not at all. The only similarity is that LLMs work with text, and the document formats can also represent text.

Each format (E.g pdf, json, excel) has a defined standard, so all you have to do to change between each other is to map one format’s fields to the others. You don’t need (and won’t get good results) from having an LLM produce the new format from scratch.

What he’s asking is the equivalent of asking if there’s an LLM made specifically for solving arithmetic problems. Why would you try to solve addition using an LLM?

It can if you don’t do a train-test split.

But even if you consider the training set only, having zero loss is definitely a bad sign.

grep file ...

There really isn’t a complicated discussion to be had unless you needlessly complicate things. There’s a big difference between having, say, better monitor or headphones in terms of resolution or sound quality vs having a monitor or headphones that add extra features.

It’s like saying that AR glasses that visualize a ball’s trajectory should be allowed in tennis or football because players can already invest in better rackets or shoes.

The detection problem is not unsolvable. First, you can forbid people that are using that monitor from matchmaking. You can find your monitor’s model number using software so that would be trivial. For a more nuanced approach, you can examine players’ reaction times and ban people that got too good too fast.