search

Jindi0 / SAFE

58 stars 15 forks source link

common issue #12

Open yanghongqi123 opened 2 years ago

yanghongqi123 commented 2 years ago

/pf_embedding/case_headline.npy’，'../pf_embedding/case_body.npy'，'../pf_embedding/case_image.npy' and '../pf_embedding/case_y_fn.npy' ? The authors have given the headline. npy and body. npy in the processed Dataset. If you want to combine the headline,you need to use show and tell tool to make case_image.npy by yourself. I tried it and used the original author's training model, but the effect was terrible and the generated text was completely incorrect. I gave up

thinking024 commented 1 year ago

Me, too. The sentence generated from images by ShowAndTell is was completely incorrect

Catrin-baze commented 1 year ago

Me, too. The sentence generated from images by ShowAndTell is was completely incorrect

How do you use show and tell to extract captions? When I use it, it seems that I can't test directly. I need the files of train and val. Are you retraining？

thinking024 commented 1 year ago

Me, too. The sentence generated from images by ShowAndTell is was completely incorrect

How do you use show and tell to extract captions? When I use it, it seems that I can't test directly. I need the files of train and val. Are you retraining？

You don't need to retraining Show and Tell, just follow the readme in it and you'll get the sentence generated from images.