Upload 7 files

app.py

@@ -28,7 +28,7 @@ st.title("Instructor XL Embeddings")
 
 
 st.write(
-    "The app compares the performance of the Instructor-XL Embedding Model on the text from AMD's Q1 2020 Earnings Call Transcript.'"
+    "The app compares the performance of the Instructor-XL Embedding Model on the text from AMD's Q1 2020 Earnings Call Transcript."
 )
 
 data = get_data()
@@ -114,29 +114,36 @@ index_mapping = {
     "Represent the earnings call transcript answer for retrieval:": "week14-instructor-xl-amd-ecta-6",
 }
 
+with col2:
+    with st.form("my_form"):
+        text_embedding_instruction = st.selectbox(
+            "Select instruction for Text Embedding",
+            text_embedding_instructions_choice,
+        )
 
-with st.form("my_form"):
-    text_embedding_instruction = st.selectbox(
-        "Select instruction for Text Embedding",
-        text_embedding_instructions_choice,
-    )
-
-    pinecone_index_name = index_mapping[text_embedding_instruction]
-    pinecone.init(
-        api_key=st.secrets[f"pinecone_{pinecone_index_name}"],
-        environment="asia-southeast1-gcp-free",
-    )
-
-    pinecone_index = pinecone.Index(pinecone_index_name)
-
-    submitted = st.form_submit_button("Submit")
-    if submitted:
-        matches = query_pinecone(
-            dense_embedding, num_results, pinecone_index, indices
+        pinecone_index_name = index_mapping[text_embedding_instruction]
+        pinecone.init(
+            api_key=st.secrets[f"pinecone_{pinecone_index_name}"],
+            environment="asia-southeast1-gcp-free",
         )
-        context = format_query(matches)
-        output_text = format_context(context)
 
+        pinecone_index = pinecone.Index(pinecone_index_name)
+
+        submitted = st.form_submit_button("Submit")
+        if submitted:
+            matches = query_pinecone(
+                dense_embedding, num_results, pinecone_index, indices
+            )
+            context = format_query(matches)
+            output_text = format_context(context)
+        
+            st.subheader("Retrieved Text:")
+            for output in output_text:
+                output = f"""{output}"""
+                st.write(
+                    f"<ul><li><p>{output}</p></li></ul>",
+                    unsafe_allow_html=True,
+                )
 
 tab1 = st.tabs(["View transcript"])
 

utils/models.py

@@ -37,19 +37,19 @@ def preprocess_text(text):
     return preprocessed_text
 
 
-@st.experimental_singleton
+@st.cache_resource
 def get_data():
     data = pd.read_csv("AMD_Q1_2020_earnings_call_data_keywords.csv")
     return data
 
 
-@st.experimental_singleton
+@st.cache_resource
 def get_instructor_embedding_model():
     client = Client("https://awinml-api-instructor-xl-1.hf.space/")
     return client
 
 
-@st.experimental_singleton
+@st.cache_resource
 def get_bm25_model(data):
     corpus = data.Text.tolist()
     corpus_clean = [preprocess_text(x) for x in corpus]