The MePO Prompt Optimization Dataset is designed to support research in prompt optimization, especially for training and evaluating MePO β a lightweight and locally deployable prompt optimization model.
rejectedThe original prompt from the Alpaca dataset, serving as the rejected example.
chosenThe optimized prompt generated by MePO, serving as the⦠See the full description on the dataset page:
https://huggingface.co/datasets/zixiaozhu/MePO_Alpaca.