search

karpathy / neuraltalk2

Efficient Image Captioning code in Torch, runs on GPU
5.51k stars 1.26k forks source link

th eval.lua questions #102

Open leo2105 opened 8 years ago

leo2105 commented 8 years ago

$th eval.lua -model model -image_folder image -num_images 1 well, i dont know what is the meaning of " -model model " ... i need examples how to run it. What kind of model are there?

cjds commented 8 years ago

Model is a t7 file that contains the learned weights. You are evaluating your network which is loaded up with the model parameter. In neuraltalk2 there is the VGG Net 16 model that is used as a base model from which you can learn weights according to your application

You can download the t7 file for VGG 16 here

Then run something like 

$th eval.lua -model model.t7 -image_folder image -num_images 1
lymrty commented 7 years ago

well, i got an question? before running the eval.lua, i must have a copy of this NeuralTalk2 project on my local, and then cd /home/myself/neuraltalk2, so i can get the location of the eval.lua file. After that , i can run this command $th eval.lua -model model.t7 -image_folder image -num_images 1??? right?