Update app.py

app.py

@@ -11,18 +11,21 @@ def summarize_function(notes):
     st.write('Summary: ')
     return gen_text
 
-notes_df = pd.read_csv('notes_small.csv')
-examples_tuple = ()
-for i in range(len(notes_df)):
-    examples_tuple += (f"Patient {i+1}", )
+@st.cache
+def notes_select():
+    notes_df = pd.read_csv('notes_small.csv')
+    examples_tuple = ()
+    for i in range(len(notes_df)):
+        examples_tuple += (f"Patient {i+1}", )    
+    return st.sidebar.selectbox('Example', (examples_tuple), index=0)
 
-example = st.sidebar.selectbox('Example', (examples_tuple), index=0)
+example = notes_select()
 
 st.markdown("<h1 style='text-align: center; color: #489DDB;'>GPT Clinical Notes Summarizer</h1>", unsafe_allow_html=True)
 st.markdown("<h6 style='text-align: center; color: #489DDB;'>by Bryan Mildort</h1>", unsafe_allow_html=True)
 
 st.sidebar.markdown("<h1 style='text-align: center; color: #489DDB;'>GPT Clinical Notes Summarizer 0.1v</h1>", unsafe_allow_html=True)
-st.sidebar.markdown("<h6 style='text-align: center; color: #489DDB;'>The model for this application was created with generous support of the Google TPU Research Cloud (TPU). This demo is for investigative research purposes only. The model is assumed to have several limiations and biases, so please oversee responses with human moderation. It is not intended for production ready enterprises and is displayed to illustrate the capabilities of Large Language Models for healthcare research.</h1>", unsafe_allow_html=True)
+st.sidebar.markdown("<h6 style='text-align: center; color: #489DDB;'>The model for this application was created with generous support of the Google TPU Research Cloud (TRC). This demo is for investigative research purposes only. The model is assumed to have several limiations and biases, so please oversee responses with human moderation. It is not intended for production ready enterprises and is displayed to illustrate the capabilities of Large Language Models for healthcare research.</h1>", unsafe_allow_html=True)
 tower = Image.open('howard_social.png')
 seal =  Image.open('Howard_University_seal.svg.png')
 st.sidebar.image(tower)
@@ -36,7 +39,7 @@ from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 # device_map = infer_auto_device_map(model, dtype="float16")
 # st.write(device_map)
 
-# @st.cache(allow_output_mutation=True)
+@st.cache(allow_output_mutation=True)
 def load_model():
     model = AutoModelForCausalLM.from_pretrained("bryanmildort/gpt_neo_notes", low_cpu_mem_usage=True)
     # model = model.to(device)