Upload app.py

app.py

@@ -5,21 +5,35 @@ from huggingface_hub import InferenceClient
 import gradio as gr
 import os
 
-print("1. Downloading Data...")
-raw_data = load_dataset("bitext/Bitext-customer-support-llm-chatbot-training-dataset")
-df = pd.DataFrame(raw_data['train'])
-df.dropna(subset=['instruction', 'response'], inplace=True)
-df.drop_duplicates(subset=['instruction'], inplace=True)
-df = df[['category', 'instruction', 'response']]
-df.columns = ['ticket_category', 'customer_query', 'historical_resolution']
+print("1. Ingesting Data (with Fallback Protection)...")
+try:
+    # Attempting to load a highly stable public dataset
+    raw_data = load_dataset("Kaludi/Customer-Support-Responses")
+    df = pd.DataFrame(raw_data['train'])
+    # Standardize the columns for our pipeline
+    df = df[['query', 'response']]
+    df.columns = ['customer_query', 'historical_resolution']
+    print("Cloud dataset loaded successfully.")
+except Exception as e:
+    print(f"Cloud dataset unavailable ({e}). Initializing backup demo dataset...")
+    # Defensive programming: Fallback data ensures your portfolio never crashes for a recruiter
+    backup_data = [
+        {"customer_query": "How do I reset my password?", "historical_resolution": "To reset your password, navigate to the login page and click 'Forgot Password'. You will receive an email with a secure reset link valid for 24 hours."},
+        {"customer_query": "Where can I find my monthly invoice?", "historical_resolution": "All invoices are securely located in your account dashboard under the 'Billing & Payments' tab. You can download them directly as PDFs."},
+        {"customer_query": "How do I cancel my current subscription?", "historical_resolution": "To cancel your subscription, go to Settings > Subscription and click 'Cancel Plan'. Your access will remain active until the end of your current billing cycle."},
+        {"customer_query": "Can I upgrade my account to the Pro tier?", "historical_resolution": "Yes, you can upgrade at any time. Go to your Account Settings, select 'Upgrade Plan', and choose the Pro tier. The prorated amount will be charged to your card immediately."},
+        {"customer_query": "My application keeps crashing on startup.", "historical_resolution": "Please ensure you are running the latest version of the app. Try clearing your browser cache. If the issue persists, please reply with your device OS and error logs."}
+    ]
+    df = pd.DataFrame(backup_data)
 
 print("2. Loading Embedding Model & Vectorizing Data...")
 model = SentenceTransformer('paraphrase-multilingual-MiniLM-L12-v2')
+# Use the available rows (up to 100 for speed)
 sample_resolutions = df['historical_resolution'].head(100).tolist()
 resolution_embeddings = model.encode(sample_resolutions)
 
 print("3. Connecting to LLM...")
-# This pulls the HF_TOKEN securely from the Spaces Settings > Secrets you set earlier
+# This pulls the HF_TOKEN securely from the Spaces Settings > Secrets
 hf_token = os.environ.get("HF_TOKEN")
 client = InferenceClient(
     model="meta-llama/Meta-Llama-3-8B-Instruct",
@@ -59,7 +73,8 @@ with gr.Blocks() as demo:
     
     with gr.Row():
         with gr.Column():
-            input_query = gr.Textbox(lines=4, label="Incoming Customer Ticket (Try German, French, etc.)")
+            # Hint updated to reflect the fallback dataset topics
+            input_query = gr.Textbox(lines=4, label="Incoming Customer Ticket (Try asking about billing or passwords in German, Spanish, etc.)")
             submit_btn = gr.Button("Resolve Ticket")
         
         with gr.Column():