fundamentally, an llm doesn’t “use” individual sources for any answer. it is just a function approximator, and as such every datapoint influences the result, just more if it closely aligns with the input.
fundamentally, an llm doesn’t “use” individual sources for any answer. it is just a function approximator, and as such every datapoint influences the result, just more if it closely aligns with the input.