How to generate the evaluation datasets?

[TURLEing]

Hi, thanks for sharing your work, it's very impressive. We'd like to evaluate the M2UGen model and other compared models with your evalution datasets. In Evaluation/Image2Music/evaluate.py, we noticed that there is a file called MUImageEvalInstructions.json, which might be the splits of evaluation dataset. However, we didn't find this file anywhere, neither in Github nor Huggingface.

Could you please share this file or sharing the code for generating the split of evaluation dataset? It will help a lot. Thanks!

[Mao-KU]

Same issue here, did you find the evaluation datasets?