kelvinxu/arctic-captions

why don't finetune cnn?

vanpersie32 opened this issue · 1 comments

hey, I have noticed that cnn-lstm model can benefit a lot from fine-tuning cnn. But why the code don't fine-tune cnn?

Hi, as we explain in the paper, this would complicate comparisons. Of course, improving the vision features would help a lot.