Update app.py

app.py

@@ -1,21 +1,26 @@
-# let's import the libraries we need
-#from sentence_transformers import SentenceTransformer
-#from sentence_transformers import CrossEncoder
+# let's import the libraries
+from email import header
+import streamlit as st
+import pandas as pd
+import numpy as np
+import pickle
 import spacy
 from sklearn.metrics.pairwise import cosine_similarity
 from datasets import load_dataset
 import io
 import netrc
-import pickle
-import sys
-import pandas as pd
-import numpy as np
-import streamlit as st
-import torch
 from tqdm import tqdm
 tqdm.pandas()
+import torch
+import os
+import sys
+import time
+import sentence_transformers
+from sentence_transformers import SentenceTransformer
+from sentence_transformers import CrossEncoder
+from sentence_transformers import util
 
-# Load the English STSB dataset
+# let's load the english stsb dataset
 stsb_dataset = load_dataset('stsb_multi_mt', 'en')
 stsb_train = pd.DataFrame(stsb_dataset['train'])
 stsb_test = pd.DataFrame(stsb_dataset['test'])
@@ -23,19 +28,16 @@ stsb_test = pd.DataFrame(stsb_dataset['test'])
 # let's create helper functions
 nlp = spacy.load("en_core_web_sm")
 
-
 def text_processing(sentence):
     sentence = [token.lemma_.lower()
                 for token in nlp(sentence)
                 if token.is_alpha and not token.is_stop]
     return sentence
 
-
 def cos_sim(sentence1_emb, sentence2_emb):
     cos_sim = cosine_similarity(sentence1_emb, sentence2_emb)
     return np.diag(cos_sim)
 
-
 # let's read the csv file
 data = (pd.read_csv("SBERT_data.csv")).drop(['Unnamed: 0'], axis=1)
 
@@ -46,61 +48,35 @@ data.rename(columns={'target_text': 'sentence2',
 data['sentence2'] = data['sentence2'].astype('str')
 data['sentence1'] = data['sentence1'].astype('str')
 
+# loop through the data
 XpathFinder = CrossEncoder("cross-encoder/stsb-roberta-base")
 sentence_pairs = []
 for sentence1, sentence2 in zip(data['sentence1'], data['sentence2']):
-    sentence_pairs.append([sentence1, sentence2])
+      sentence_pairs.append([sentence1, sentence2])
 
 data['SBERT CrossEncoder_Score'] = XpathFinder.predict(
-    sentence_pairs, show_progress_bar=True)
-
-# sorting the values
-data.sort_values(by=['SBERT CrossEncoder_Score'], ascending=False)
+      sentence_pairs, show_progress_bar=True)
 
 loaded_model = XpathFinder
 
-# Containers
+# let's create containers
 header_container = st.container()
 mod_container = st.container()
 
-# Header
+# let's create the header
 with header_container:
+      st.title("SBERT CrossEncoder")
+      st.markdown("This is a demo of the SBERT CrossEncoder model")
 
-    # different levels of text you can include in your app
-    st.title("Xpath Finder App")
-
-
-# model container
+# let's create the model container
 with mod_container:
-
-    # collecting input from user
-    prompt = st.text_input("Enter your description below ...")
-
-    # Loading e data
-    data = (pd.read_csv("/content/SBERT_data.csv")
-            ).drop(['Unnamed: 0'], axis=1)
-
-    data['prompt'] = prompt
-    data.rename(columns={'target_text': 'sentence2',
-                'prompt': 'sentence1'}, inplace=True)
-    data['sentence2'] = data['sentence2'].astype('str')
-    data['sentence1'] = data['sentence1'].astype('str')
-
-    # let's pass the input to the loaded_model with torch compiled with cuda
-    if prompt:
-        # let's get the result
-        simscore = loaded_model.predict([prompt])
-
-        from sentence_transformers import CrossEncoder
-        loaded_model = CrossEncoder("cross-encoder/stsb-roberta-base")
-        sentence_pairs = []
-        for sentence1, sentence2 in zip(data['sentence1'], data['sentence2']):
-            sentence_pairs.append([sentence1, sentence2])
-
-        # sorting the df to get highest scoring xpath_container
-        data['SBERT CrossEncoder_Score'] = loaded_model.predict(sentence_pairs)
-        most_acc = data.head(5)
-        # predictions
-        st.write("Highest Similarity score: ", simscore)
-        st.text("Is this one of these the Xpath you're looking for?")
-        st.write(st.write(most_acc["input_text"]))
+   # let's get input from the user
+   prompt = st.text_input("Enter a description below...")
+
+   if prompt:
+      simscore = loaded_model.predict([prompt])
+      # sort the values
+      data['SBERT CrossEncoder_Score'] = simscore
+      most_acc = data.head(5)
+      st.write(most_acc)
+      st.write("The most accurate sentence is: ", most_acc['sentence2'].iloc[0])