I plan to fine tune davinci-002 using the legacy fine tuning API.
Do I need to initially shuffle the dataset (that must be shuffled for normal fine tuning) before submitting
.jsonl file to fine tuning API? Or, do the API automatically and randomly sample the data from the dataset to construct a batch for each step?
(I strongly suspect that they would use random sampler for the training, but I found that some examples in OpenAI cookbook shuffles their dataset before submitting the jsonl file to the API.)