add cache

app.py

@@ -198,7 +198,7 @@ def topic_distribution(weights, smoothing=0.01):
  # return(topic_frequencies[category], topic_frequencies_spotlight[category], topic_ratios[category], category)
 
 def populate_session(dataset,model):
- data_df = pd.read_parquet('./assets/data/'+dataset+ '_'+ model+'.parquet')
+ data_df = read_file_to_df('./assets/data/'+dataset+ '_'+ model+'.parquet')
  if model == 'albert-base-v2-yelp-polarity':
  tokenizer = AutoTokenizer.from_pretrained('textattack/'+model)
  else:
@@ -208,7 +208,9 @@ def populate_session(dataset,model):
  if "selected_slice" not in st.session_state:
  st.session_state["selected_slice"] = None
  
-
+@st.cache(ttl=600)
+def read_file_to_df(file):
+ return pd.read_parquet(file)
 
 if __name__ == "__main__":
  ### STREAMLIT APP CONGFIG ###
@@ -235,7 +237,7 @@ if __name__ == "__main__":
  ### LOAD DATA AND SESSION VARIABLES ###
  ##uncomment the next next line to run dynamically and not from file
  #populate_session(dataset, model)
- data_df = pd.read_parquet('./assets/data/'+dataset+ '_'+ model+'.parquet')
+ data_df = read_file_to_df('./assets/data/'+dataset+ '_'+ model+'.parquet')
  loss_quantile = st.sidebar.slider(
  "Loss Quantile", min_value=0.5, max_value=1.0,step=0.01,value=0.95
  )
@@ -250,7 +252,7 @@ if __name__ == "__main__":
  st.markdown('<h3>Word Distribution in Error Slice</h3>', unsafe_allow_html=True)
  #uncomment the next two lines to run dynamically and not from file
  #commontokens = frequent_tokens(data_df, tokenizer, loss_quantile=loss_quantile)
- commontokens = pd.read_parquet('./assets/data/'+dataset+ '_'+ model+'_commontokens.parquet')
+ commontokens = read_file_to_df('./assets/data/'+dataset+ '_'+ model+'_commontokens.parquet')
  with st.expander("How to read the table:"):
  st.markdown("* The table displays the most frequent tokens in error slices, relative to their frequencies in the val set.")
  st.write(commontokens)
@@ -260,20 +262,22 @@ if __name__ == "__main__":
  num_clusters = st.sidebar.slider("# clusters", min_value=1, max_value=20, step=1, value=3)
 
  if run_kmeans == 'True':
- merged = kmeans(data_df,num_clusters=num_clusters)
+ with st.spinner(text='running kmeans...'):
+ merged = kmeans(data_df,num_clusters=num_clusters)
  with lcol:
  st.markdown('<h3>Error Slices</h3>',unsafe_allow_html=True)
- dataframe=pd.read_parquet('./assets/data/'+dataset+ '_'+ model+'_error-slices.parquet')
+ with st.expander("How to read the table:"):
+ st.markdown("* *Error slice* refers to the subset of evaluation dataset the model performs poorly on.")
+ st.markdown("* The table displays model error slices on the evaluation dataset, sorted by loss.")
+ st.markdown("* Each row is an input example that includes the label, model pred, loss, and error cluster.")
+ with st.spinner(text='loading error slice...'):
+ dataframe=read_file_to_df('./assets/data/'+dataset+ '_'+ model+'_error-slices.parquet')
  #uncomment the next next line to run dynamically and not from file
  # dataframe = merged[['content', 'label', 'pred', 'loss', 'cluster']].sort_values(
  # by=['loss'], ascending=False)
  # table_html = dataframe.to_html(
  # columns=['content', 'label', 'pred', 'loss', 'cluster'], max_rows=50)
  # table_html = table_html.replace("<th>", '<th align="left">') # left-align the headers
- with st.expander("How to read the table:"):
- st.markdown("* *Error slice* refers to the subset of evaluation dataset the model performs poorly on.")
- st.markdown("* The table displays model error slices on the evaluation dataset, sorted by loss.")
- st.markdown("* Each row is an input example that includes the label, model pred, loss, and error cluster.")
  st.write(dataframe,width=900, height=300)
 
  quant_panel(merged)