by doodlewind on 3/24/25, 7:26 AM with 2 comments
by nchagnet on 3/24/25, 9:28 AM
This is really interesting! You mention using LLMs to parse sentences in strongly typed prompts, but how would this work with the prevalence of tolerated grammatical errors in most languages? Would such a parsing always be "pass/fail" or would you be able to get a degree of correctness?