I am trying to reproduce the retrieval task results for Flores dataset of IndicXtreme. I am using. - fine-tuning/retrieval/retrieval.py code.
However, the accuracy returned by the script is nowhere closer to what is reported in the papers. Please help me understand.
I am trying to reproduce the retrieval task results for Flores dataset of IndicXtreme. I am using. - fine-tuning/retrieval/retrieval.py code. However, the accuracy returned by the script is nowhere closer to what is reported in the papers. Please help me understand.