BARTScore
BARTScore copied to clipboard
How to choose scoring model?
The evaluation scores seems to be easily affected by the selection of fine-tuned models on different datasets