Update app.py

app.py

@@ -2,23 +2,19 @@ import streamlit as st
 import pandas as pd
 import json
 import joblib
-import os
+import torch
 
-# Function to load models from local files
+# Function to load model safely on CPU
 def load_model(model_path):
-    """Loads the model from the given local path."""
-    return joblib.load(model_path)
+    """Loads the model, ensuring it is mapped to the CPU."""
+    return torch.load(model_path, map_location=torch.device('cpu'))
 
-# Paths to your uploaded models in Hugging Face
-model_folder = 'models'
-if not os.path.exists(model_folder):
-    os.makedirs(model_folder)
+# Model file paths (ensure they are uploaded in the correct directory)
+distilbert_model_path = "models/distilbert_model.joblib"
+bert_topic_model_path = "models/bertopic_model.joblib"
+recommendation_model_path = "models/recommendation_model.joblib"
 
-distilbert_model_path = os.path.join(model_folder, 'distilbert_model.joblib')
-bert_topic_model_path = os.path.join(model_folder, 'bertopic_model.joblib')
-recommendation_model_path = os.path.join(model_folder, 'recommendation_model.joblib')
-
-# Load models
+# Load all models with CPU mapping
 distilbert_model = load_model(distilbert_model_path)
 bert_topic_model = load_model(bert_topic_model_path)
 recommendation_model = load_model(recommendation_model_path)
@@ -40,11 +36,7 @@ def extract_feedback(file):
         return feedback_text
     elif file.type == "application/json":
         json_data = json.load(file)
-        feedback_text = []
-        if isinstance(json_data, list):
-            feedback_text = [item.get('feedback', '') for item in json_data if 'feedback' in item]
-        elif isinstance(json_data, dict):
-            feedback_text = list(json_data.values())
+        feedback_text = [item.get('feedback', '') for item in json_data if isinstance(item, dict)]
         return feedback_text
     elif file.type == "text/plain":
         return [file.getvalue().decode("utf-8")]
@@ -54,19 +46,17 @@ def extract_feedback(file):
 # Display error or feedback extraction status
 if uploaded_file:
     feedback_text_list = extract_feedback(uploaded_file)
+
     if feedback_text_list:
         for feedback_text in feedback_text_list:
             if st.button(f'Analyze Feedback: "{feedback_text[:30]}..."'):
-                # Sentiment Analysis
                 sentiment = distilbert_model.predict([feedback_text])
                 sentiment_result = 'Positive' if sentiment == 1 else 'Negative'
                 st.write(f"Sentiment: {sentiment_result}")
-                
-                # Topic Modeling
+
                 topics = bert_topic_model.predict([feedback_text])
                 st.write(f"Predicted Topic(s): {topics}")
-                
-                # Recommendation System
+
                 recommendations = recommendation_model.predict([feedback_text])
                 st.write(f"Recommended Actions: {recommendations}")
     else: