Update app.py

app.py

@@ -196,29 +196,6 @@ class Search:
             similarity_scores[choice] = similarity
         return similarity_scores
 
-"""## 3. Word Arithmetic
-
-Let's test your embeddings. Answer the question below through the search functionality you implemented above
-"""
-
-embeddings_model = Embeddings()
-search_using_cos = Search(embeddings_model)
-
-word_index_dict, embeddings = embeddings_model.load_glove_embeddings(50)
-
-current_embedding = embeddings_model.embeddings_preprocess( word_index_dict, ["king", "woman"], ["man"], embeddings)
-
-closest_word = search_using_cos.find_closest_words(current_embedding, ["girl", "queen", "princess", "daughter", "mother"], word_index_dict, embeddings )
-
-print("'King - Man + Woman':", closest_word)
-
-
-word_index_dict, embeddings = embeddings_model.load_glove_embeddings(50)
-
-
-closest_word = search_using_cos.find_word_as( ("tesla", "car"), "apple", ["fruit", "vegetable", "gas"], word_index_dict, embeddings)
-
-print("'Tesla:Car as Apple:?': ", closest_word)
 
 """## 4. Plots
 
@@ -238,43 +215,9 @@ def plot_pie_chart(category_similarity_scores):
     ax.axis('equal')  # Equal aspect ratio ensures the pie chart is circular.
     plt.show()
 
-word_index_dict, embeddings = embeddings_model.load_glove_embeddings(50)
-
-# Find the word closest to the vector resulting from "king" - "man" + "woman"
-current_embedding = embeddings_model.embeddings_preprocess(word_index_dict, ["king", "woman"], ["man"], embeddings)
-
-# Calculate similarity scores for a set of words and plot them
-sim_scores = search_using_cos.find_similarity_scores(current_embedding, ["girl", "queen", "princess", "daughter", "mother"], word_index_dict, embeddings)
-plot_pie_chart(sim_scores)
-
-"""## 5. Test
-
-Test your pie chart against some of the examples in the demo listed here:
-
-https://categorysearch.streamlit.app or
-https://searchdemo.streamlit.app
 
-a) Do the results make sense?
-b) Which embedding gives more meaningful results?
 
-"""
-
-input_sentence = "Roses are red, trucks are blue, and Seattle is grey right now"
-category_names = ["Flowers", "Colors", "Cars", "Weather", "Food"]
-
-embeddings_model = Embeddings()
-word_index_dict, embeddings = embeddings_model.load_glove_embeddings(50)
-categories_embedding = {category: embeddings_model.get_sentence_transformer_embedding(category) for category in category_names}
-
-search_instance = Search(embeddings_model)
-category_similarity_scores = search_instance.get_topK_similar_categories(input_sentence, categories_embedding)
 
-plot_pie_chart(category_similarity_scores) # Plot and see
-
-"""## 6. Bonus (if time permits)!
-Create a simple streamlit or equivalent webapp like the link in 5.
-This is also part of your Mini-Project 1!
-"""
 
 def plot_piechart(sorted_cosine_scores_items):
     sorted_cosine_scores = np.array([
@@ -365,67 +308,41 @@ def plot_alatirchart(sorted_cosine_scores_models):
 
 
 ### Text Search ###
-st.sidebar.title("GloVe Twitter")
-st.sidebar.markdown(
-    """
-GloVe is an unsupervised learning algorithm for obtaining vector representations for words. Pretrained on
-2 billion tweets with vocabulary size of 1.2 million. Download from [Stanford NLP](http://nlp.stanford.edu/data/glove.twitter.27B.zip).
+st.sidebar.title("sentence transformer")
 
-Jeffrey Pennington, Richard Socher, and Christopher D. Manning. 2014. *GloVe: Global Vectors for Word Representation*.
-"""
-)
 
-# initialize Session State variable
 if 'categories' not in st.session_state:
     st.session_state['categories'] = "Flowers Colors Cars Weather Food"
 if 'text_search' not in st.session_state:
     st.session_state['text_search'] = "Roses are red, trucks are blue, and Seattle is grey right now"
 
-model_type = st.sidebar.selectbox("Choose the model", ("25d", "50d"), index=1)
+embeddings_model = Embeddings()
 
+model_type = st.sidebar(("50d"), index=1)
 
-st.title("In Class practice 1 demo")
+st.title("in in-class coding practice1 Demo")
 st.subheader(
     "Pass in space separated categories you want this search demo to be about."
 )
-# st.selectbox(label="Pick the categories you want this search demo to be about...",
-# options=("Flowers Colors Cars Weather Food", "Chocolate Milk", "Anger Joy Sad Frustration Worry Happiness", "Positive Negative"),
-# key="categories"
-# )
-
 
 # categories of user input
-categories = st.text_input(
+user_categories = st.text_input(
     label="Categories", value=st.session_state.categories
 )
 
-st.session_state.categories = categories.split(" ")
+st.session_state.categories = user_categories.split(" ")
 
 print(st.session_state.get("categories"))
+
 print(type(st.session_state.get("categories")))
-# print("Categories = ", categories)
-# st.session_state.categories = categories
 
 st.subheader("Pass in an input word or even a sentence")
-text_search = st.text_input(
+user_text_search = st.text_input(
     label="Input your sentence",
-    st.session_state.text_search,
+    value=st.session_state.text_search,
 )
 
-st.session_state.text_search = text_search
-
-# Download glove embeddings if it doesn't exist
-embeddings_path = "embeddings_" + str(model_type) + "_temp.npy"
-word_index_dict_path = "word_index_dict_" + str(model_type) + "_temp.pkl"
-if not os.path.isfile(embeddings_path) or not os.path.isfile(word_index_dict_path):
-    print("Model type = ", model_type)
-    glove_path = "Data/glove_" + str(model_type) + ".pkl"
-    print("glove_path = ", glove_path)
-
-    # Download embeddings from google drive
-    with st.spinner("Downloading glove embeddings..."):
-        download_glove_embeddings_gdrive(model_type)
-
+st.session_state.text_search = user_text_search
 
 # Load glove embeddings
 word_index_dict, embeddings = embeddings_model.load_glove_embeddings(model_type)
@@ -436,8 +353,8 @@ category_embeddings = {category: embeddings_model.get_sentence_transformer_embed
 search_using_cos = Search(embeddings_model)
 
 # Find closest word to an input word
-if st.session_state.get("text_search"):
-    # sentence transformer  Embedding
+if st.session_state.text_search:
+    # sentence transformer embeddings
     print("sentence transformer  Embedding")
     embeddings_metadata = {
         "word_index_dict": word_index_dict,
@@ -445,18 +362,16 @@ if st.session_state.get("text_search"):
         "model_type": model_type,
         "text_search": st.session_state.text_search
     }
-    with st.spinner("Obtaining Cosine similarity ..."):
+    with st.spinner("Obtaining Cosine similarity for Glove..."):
         sorted_cosine_sim_transformer = search_using_cos.get_topK_similar_categories(
             st.session_state.text_search, category_embeddings
         )
 
-    
-
     # Results and Plot Pie Chart for Glove
     print("Categories are: ", st.session_state.categories)
     st.subheader(
         "Closest word I have between: "
-        + st.session_state.categories
+        + " ".join(st.session_state.categories)
         + " as per different Embeddings"
     )
 
@@ -466,13 +381,13 @@ if st.session_state.get("text_search"):
 
     st.write(
         f"Closest category using sentence transformer embeddings : {list(sorted_cosine_sim_transformer.keys())[0]}")
-    
+
     plot_alatirchart(
         {
             "sentence_transformer_384": sorted_cosine_sim_transformer,
         }
     )
-   
+
 
     st.write("")
     st.write(