Github url added along with urls of base model used

README.md

@@ -1,3 +1,13 @@
+
+To build the model I used Resnet18 for image part and Turkish-DistillBert for text part.
+Turkish-DistillBert: [dbmdz/distilbert-base-turkish-cased]
+
+You can get more information (and code :tada:) on how to train or use the model on my [github].
+
+[dbmdz/distilbert-base-turkish-cased]: https://huggingface.co/dbmdz/distilbert-base-turkish-cased
+
+[github]: https://github.com/kesimeg/turkish-clip
+
 # How to use the model?
 
 In order to use the model use can use the class in model.py like the example below:
@@ -14,7 +24,7 @@ from transformers import AutoTokenizer, AutoModel
 
 model = Net()
 # If you use model on cpu you need the map_location part
-model.load_state_dict(torch.load("clip_model.pt", map_location=torch.device('cpu'))) 
+model.load_state_dict(torch.load("clip_model.pt", map_location=torch.device('cpu')))
 model.eval()
 
 tokenizer = AutoTokenizer.from_pretrained("dbmdz/distilbert-base-turkish-cased")
@@ -35,7 +45,7 @@ def predict(img,text_vec):
   mask = torch.Tensor(token_list["attention_mask"]).long()
 
 
-  image_vec, text_vec = model(input, text , mask) 
+  image_vec, text_vec = model(input, text , mask)
   print(F.softmax(torch.matmul(image_vec,text_vec.T),dim=1))
 
 img = Image.open("dog.png") # A dog image
@@ -44,5 +54,3 @@ text_vec = ["Çimenler içinde bir köpek.","Bir köpek.","Çimenler içinde bir
 predict(img,text_vec) # Probabilities for each description
 
 ```
-
-