easy5 pts · first answer countsllm-fundamentalsllm-basics

Pretraining data compared with tuning data

Which comparison of data size is usually right?

A.Fine-tuning uses trillions of tokens; pretraining uses a fewB.Pretraining uses trillions of tokens; fine-tuning uses far lessC.Both always use exactly the same dataset and sizeD.Pretraining uses only labelled question-answer pairs

Discussion 0

Join free to post in the discussion.

Pretraining data compared with tuning data

Discussion 0

Related questions