Update app.py

app.py

@@ -1,11 +1,9 @@
 import streamlit as st
-from sentence_transformers import SentenceTransformer
+from sentence_transformers import SentenceTransformer, InputExample, losses
 from sklearn.metrics.pairwise import cosine_similarity
 from keyphrasetransformer import KeyPhraseTransformer
 from datasets import load_dataset
-from sentence_transformers import losses
 from torch.utils.data import DataLoader
-from sentence_transformers import InputExample
 import torch
 
 kp = KeyPhraseTransformer()
@@ -24,6 +22,13 @@ def generate_wordcloud(text, title):
     # Your existing code for generating word clouds
     pass
 
+class CustomCollate:
+    def __init__(self):
+        pass
+
+    def __call__(self, batch):
+        return batch
+
 st.title("Resume Match Calculator")
 
 model = load_model()
@@ -48,13 +53,6 @@ for i in range(n_examples):
 # Now train_examples contains InputExample instances with 'Resume' as text
 # You can use train_examples for training your sentence embedding model
 
-class CustomCollate:
-    def __init__(self):
-        pass
-
-    def __call__(self, batch):
-        return batch
-
 # Create a DataLoader for training examples with custom collate function
 batch_size = 16
 train_dataloader = DataLoader(train_examples, shuffle=True, batch_size=batch_size, collate_fn=CustomCollate())
@@ -73,7 +71,7 @@ for epoch in range(epochs):
 
     for batch in train_dataloader:
         optimizer.zero_grad()
-        embeddings = model.encode(batch['texts'])
+        embeddings = model.encode(batch[0]['texts'])
         loss_value = train_loss.compute_loss(embeddings, torch.zeros_like(embeddings))
         loss_value.backward()
         optimizer.step()
@@ -119,5 +117,3 @@ if st.button("Calculate Match Score"):
 
     else:
         st.write("Please enter both the job description and resume.")
-
-