app.py

app.py

@@ -0,0 +1,74 @@
+import time
+import gradio as gr
+
+# -------------------------------
+# Simulated QA Models
+# -------------------------------
+def qa_system(method, question):
+    start_time = time.time()
+
+    if not question.strip():
+        return "**Error:** Please enter a question.", 0.0, "0 seconds", ""
+
+    # Simulated response based on method
+    if method == "Retrieval-Augmented Generation (RAG)":
+        answer = "Using RAG: Based on retrieved financial documents, the answer is $95,000,000."
+        model_name = "RAG-based Model"
+        confidence = 0.92
+    else:
+        # Fine-tuned TinyLLaMA + LoRA response simulation
+        answer = "The total revenue in 2023 was $100,000,000."
+        model_name = "Fine-Tuned TinyLLaMA-1.1B (LoRA)"
+        confidence = 0.95
+
+    end_time = time.time()
+    response_time = round(end_time - start_time, 2)
+
+    return (
+        f"**Method:** {model_name}",
+        confidence,
+        f"{response_time} seconds",
+        answer
+    )
+
+# -------------------------------
+# Gradio UI
+# -------------------------------
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown(
+        """
+        # 📊 Comparative Financial QA System  
+        Compare **Retrieval-Augmented Generation (RAG)**  
+        and **Fine-Tuned TinyLLaMA LoRA** models for Microsoft's financial Q&A.
+        """
+    )
+
+    with gr.Row():
+        method = gr.Radio(
+            choices=["Retrieval-Augmented Generation (RAG)", "Fine-Tuned Model"],
+            label="Choose QA Method:",
+            value="Fine-Tuned Model"
+        )
+
+    question = gr.Textbox(
+        label="Ask a question about Microsoft's 2022-2023 financials:",
+        placeholder="e.g., What was the total revenue in 2023?"
+    )
+
+    submit_btn = gr.Button("Get Answer")
+
+    # Output section
+    method_output = gr.Markdown()
+    confidence_output = gr.Number(label="Model Confidence")
+    response_time_output = gr.Textbox(label="Response Time")
+    answer_output = gr.Markdown(label="Answer")
+
+    submit_btn.click(
+        qa_system,
+        inputs=[method, question],
+        outputs=[method_output, confidence_output, response_time_output, answer_output]
+    )
+
+# -------------------------------
+# Launch for Hugging Face Spaces
+demo.launch()