Update app.py

app.py

@@ -4,12 +4,12 @@ import torch
 from transformers import BertTokenizerFast, BertForSequenceClassification
 import os
 
-app = Flask(__name__)
+# Configure Flask to serve static files from 'dist' folder
+app = Flask(__name__, static_folder='dist', static_url_path='/')
 CORS(app)
 
 # Set the device
-device = "cpu" # HF Spaces free tier is CPU only usually, or we can check.
-# But for safety and simplicity on free tier, let's stick to CPU or auto-detect but expect CPU.
+device = "cpu" 
 if torch.cuda.is_available():
     device = "cuda"
 elif torch.backends.mps.is_available():
@@ -20,11 +20,6 @@ else:
 print(f"Using device: {device}")
 
 # Load the saved model and tokenizer
-# In HF Spaces, we will upload the model files to the root of the repository or a subfolder.
-# Let's assume the user uploads the contents of 'saved_bert_sentiment_model3final' to the root of the Space.
-# So the path is just "." or specific folder if they upload the folder itself.
-# To be safe and organized, let's assume they upload the folder 'model' or just files to root.
-# Best practice for manual upload: Upload files to root.
 MODEL_PATH = "." 
 
 try:
@@ -36,11 +31,16 @@ try:
     print("Model loaded successfully!")
 except Exception as e:
     print(f"Error loading model: {e}")
-    # In a Space, if model fails, the app might crash or just not work.
 
 @app.route('/')
 def home():
-    return "Sentiment Analysis API is Running!"
+    # Serve the React app
+    return app.send_static_file('index.html')
+
+@app.route('/<path:path>')
+def static_proxy(path):
+    # Serve other static files (JS, CSS, etc.)
+    return app.send_static_file(path)
 
 @app.route('/predict', methods=['POST'])
 def predict():
@@ -65,10 +65,8 @@ def predict():
             logits = loaded_model(input_ids, attention_mask=attention_mask).logits
         
         prediction = torch.argmax(logits, dim=1).item()
-        # 1 is Positive, 0 is Negative based on the notebook
         sentiment = "Positive" if prediction == 1 else "Negative"
         
-        # Calculate confidence (softmax)
         probs = torch.nn.functional.softmax(logits, dim=1)
         confidence = probs[0][prediction].item()
 
@@ -81,5 +79,4 @@ def predict():
         return jsonify({'error': str(e)}), 500
 
 if __name__ == '__main__':
-    # HF Spaces expects the app to run on port 7860
     app.run(debug=False, host='0.0.0.0', port=7860)