diff --git a/code/pages/2_HMM-GLM.py b/code/pages/2_HMM-GLM.py
index 8e2c67f..f3e0ac5 100644
--- a/code/pages/2_HMM-GLM.py
+++ b/code/pages/2_HMM-GLM.py
@@ -4,6 +4,7 @@
 """
 import os
 import re
+import numpy as np
 from PIL import Image
 
 import streamlit as st
@@ -42,6 +43,9 @@ def extract_session_number(file_name):
     match = re.search(r'sess_(\d+)', file_name)
     return int(match.group(1)) if match else None
 
+def extract_state_part_number(file_name):
+    match = re.search(r'state-(\d+)_part-(\d+)', file_name)
+    return int(match.group(1)), int(match.group(2)) if match else None
 
 # --------------- Main app -------------------
 def app():
@@ -111,23 +115,50 @@ def app():
                                                 
                 # Grouped by selection
                 with st.container(height=height):
-                    cols = st.columns([1.5, 1, 4])
+                    cols = st.columns([2, 1, 4])
                     grouped_by = cols[0].selectbox('Grouped By', ['grouped_by_sessions', 'grouped_by_sessions_conventional_view', 'grouped_by_states'])
-                    num_cols = cols[1].number_input(
-                        label='number of columns',
-                        min_value=1,
-                        max_value=10,
-                        value=2,
-                    )
+
+                    fig_sessions = s3.glob(f'{num_states_folder}/{grouped_by}/*.png')
                     
-                    if grouped_by:
-                        fig_sessions = s3.glob(f'{num_states_folder}/{grouped_by}/*.png')
-                        fig_sessions = sorted(fig_sessions, key=extract_session_number)
+                    if 'by_sessions' in grouped_by:
+                        # Choose number of columns
+                        num_cols = cols[1].number_input(
+                            label='number of columns',
+                            min_value=1,
+                            max_value=10,
+                            value=2,
+                        )
+
+                        # Show plots
                         cols = st.columns(num_cols)
-                        
+                        fig_sessions = sorted(fig_sessions, key=extract_session_number)
                         for i, fig_session in enumerate(fig_sessions):
                             with cols[i % num_cols]:
                                 img = open_image_from_s3(fig_session, caption=fig_session)
+                                
+                    elif 'by_states' in grouped_by:
+                        # Interpret file names
+                        state_part = np.array([extract_state_part_number(f) 
+                                               for f in fig_sessions])
+                        unique_states = np.unique(state_part[:, 0])
+                        
+                        # Select states to compare
+                        selected_state_to_compare = cols[2].multiselect('States to compare',
+                                                                   options=unique_states,
+                                                                   default=unique_states,
+                                                                   )
+                        
+                        # Show plots
+                        cols_state = st.columns(len(selected_state_to_compare))
+                        for n, state in enumerate(selected_state_to_compare):
+                            with cols_state[n]:
+                                st.markdown(f'State {state}')
+                                ids_this_state = np.where(state_part[:, 0] == state)[0]
+                                for id in ids_this_state:
+                                    open_image_from_s3(
+                                        fig_sessions[id],
+                                        caption='',
+                                        )
 
 
 app()
\ No newline at end of file