cannot get good results with the pretrained model. Is there someone could provided your model to me?

DeepRNN / image_captioning

Tensorflow implementation of "Show, Attend and Tell: Neural Image Caption Generation with Visual Attention"

MIT License

781 stars 351 forks source link

cannot get good results with the pretrained model. Is there someone could provided your model to me? #59

Closed chansongoal closed 5 years ago

chansongoal commented 5 years ago

I have loaded the pretrained model provided by the author to evaluate the "COCO val2014" dataset. But I got strange results. And the results of four metrics are lower than the ones claimed by the author. Is there someone encounter the same problem with me?