Question about FVQA and LiveVQA evaluation splits used in OpenSearch-VL

[wang-siqi7]

Hi OpenSearch-VL authors,
Thanks for releasing this excellent work.
I am trying to reproduce the evaluation results of OpenSearch-VL and would like to clarify the exact evaluation samples used in the experiments.Could you please provide the sample lists or split files used for the following benchmarks?

1. FVQA evaluation:
   • Which exact FVQA test samples were used for the reported results?
   • Is the evaluation set the same as the FVQA test split from MMSearch-R1, or another split?
2. LiveVQA evaluation:
   • Which exact LiveVQA samples were used for evaluation?
   • Since there are different versions/splits of LiveVQA, could you please specify the benchmark version and provide the corresponding sample IDs if possible?
     Having the exact evaluation sample IDs would help ensure a faithful reproduction and facilitate comparison with future work.
     Thank you very much for your help.