Evaluation Scripts

[jcwang0602]

I really appreciate your excellent work. However, the script you uploaded for the evaluation seems to have some issues. At the moment, I am unable to properly evaluate the model. The error message is as follows:

[Dyb3438]

@jcwang0602 Please check whether the data path in Line 73 points to the correct file: PaDT-MLLM/RefCOCO/refcoco_val.json.

If you have already downloaded it locally, you can specify the path like this:

data_files = [ '/home/xxx/dataset/PaDT-MLLM/RefCOCO/refcoco_val.json' ]

You can also print out item to show me what data you are currently using.

[jcwang0602]

This is the refcoco_val.json, it not include the item 'objects'

[Dyb3438]

@jcwang0602 Not this one. This file is downloaded from VLM-R1 paper, which is the path written in Line124. The data path in Line 73 should be pointed to this file.

Sorry for any inconvenience caused.

[jcwang0602]

Thank you very much.