Well was implement an ConvNeXt on a shoes classification model
without TTA , the error was TensorBase(0.0760)
However after using TTA the error was TensorBase(0.0800)
What might be the reason ? and should I now proceed with TTA being implemented with the model?