Update app.py

app.py

@@ -3,7 +3,7 @@ from transformers import AutoTokenizer, AutoModelForSequenceClassification
 import shap
 import torch
 import numpy as np
-import os
+import matplotlib.pyplot as plt
 
 # Load model and tokenizer with caching
 @st.cache_resource
@@ -71,9 +71,9 @@ if st.button("Analyze Sentiment"):
             # Generate SHAP explanations
             st.subheader("🔍 Explanation")
             st.markdown("""
-            **Word impacts**  
-            Red → Increases score | Blue → Decreases score  
-            Intensity shows magnitude of impact
+            **Feature importance (word-level impacts)**  
+            🔴 Higher positive values → Increases sentiment  
+            🔵 Lower negative values → Decreases sentiment
             """)
 
             shap_values = explainer([text_input])
@@ -82,19 +82,13 @@ if st.button("Analyze Sentiment"):
             tabs = st.tabs(output_names)
             for i, tab in enumerate(tabs):
                 with tab:
-                    # Save SHAP text explanation as an HTML file
-                    html_path = f"shap_explanation_{i}.html"
-                    with open(html_path, "w", encoding="utf-8") as file:
-                        file.write(shap.plots.text(shap_values[:, :, i], display=False))
+                    # Create a bar plot of SHAP values
+                    fig, ax = plt.subplots(figsize=(8, 4))
+                    shap.plots.bar(shap_values[:, :, i], show=False)
                     
-                    # Read and display the saved HTML file in Streamlit
-                    with open(html_path, "r", encoding="utf-8") as file:
-                        shap_html = file.read()
-                    
-                    st.components.v1.html(shap_html, height=400, scrolling=True)
-
-                    # Clean up temporary HTML files (optional)
-                    os.remove(html_path)
+                    # Display the plot in Streamlit
+                    st.pyplot(fig)
+                    plt.close(fig)  # Free memory after rendering
 
     else:
         st.warning("Please enter some text to analyze")