Update app.py

app.py

@@ -2,17 +2,11 @@ import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
 # --- Model Loading ---
-# It's recommended to load the model and tokenizer once outside the function
-# to avoid reloading them on every call.
 model_id = "HAissa/EdNA"
 model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 
 def get_answer(question, ans_a, ans_b, ans_c, ans_d):
-    """
-    This function takes a question and four answers, formats them for the EdNA model,
-    and returns the model's predicted answer.
-    """
     options = f"A) {ans_a}\nB) {ans_b}\nC) {ans_c}\nD) {ans_d}"
     prompt = f"Question: {question}\nOptions:\n{options}\nAnswer:"
 
@@ -22,7 +16,6 @@ def get_answer(question, ans_a, ans_b, ans_c, ans_d):
     outputs = model.generate(**inputs, max_new_tokens=3)
     answer_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
-    # Parse the final answer
     try:
         final_answer = answer_text.split("Answer:")[1].strip().split('\n')[0]
         if final_answer.startswith("A)"):
@@ -35,7 +28,6 @@ def get_answer(question, ans_a, ans_b, ans_c, ans_d):
             return ans_d
         else:
             return final_answer
-        
     except IndexError:
         final_answer = "Could not parse the model's answer."
 
@@ -43,26 +35,50 @@ def get_answer(question, ans_a, ans_b, ans_c, ans_d):
 
 # --- Gradio Interface ---
 
-# Define a soft theme with a green primary color and a custom font for a more modern look
+# Use a modern font (Poppins) and 'emerald' for a fresh green look.
+# Increased radius_size gives components a friendlier, modern rounded look.
 theme = gr.themes.Soft(
-    primary_hue="green",
-    font=[gr.themes.GoogleFont("Karla"), "Arial", "sans-serif"]
+    primary_hue="emerald",
+    font=[gr.themes.GoogleFont("Poppins"), "ui-sans-serif", "system-ui", "sans-serif"],
+    radius_size="lg"
 )
 
 with gr.Blocks(theme=theme) as demo:
-    gr.Markdown("# EdNA: MCQ Answering App")
-    gr.Markdown("Enter a multiple-choice question and four possible answers, then click 'Get Answer' to see the model's prediction.")
+    gr.Markdown(
+        """
+        # 🤖 EdNA: MCQ Answering AI
+        Enter your question and options below, then click **Predict Answer** to see the model's choice.
+        """
+    )
+
+    with gr.Group():
+        # Question input: larger, but limited max_lines to prevent excessive scrolling
+        question_input = gr.Textbox(
+            label="Question",
+            placeholder="Type the full question here...",
+            lines=3,
+            max_lines=5
+        )
 
-    with gr.Column():
-        question_input = gr.Textbox(label="Question", placeholder="Enter your question here...")
-        answer_a_input = gr.Textbox(label="Answer A", placeholder="Enter answer A")
-        answer_b_input = gr.Textbox(label="Answer B", placeholder="Enter answer B")
-        answer_c_input = gr.Textbox(label="Answer C", placeholder="Enter answer C")
-        answer_d_input = gr.Textbox(label="Answer D", placeholder="Enter answer D")
+        # 2x2 Grid for a compact, modern MCQ layout
+        with gr.Row():
+            # setting max_lines=1 ensures these stay as single-line, non-scrollable input fields
+            answer_a_input = gr.Textbox(label="Option A", placeholder="Answer A", lines=1, max_lines=1)
+            answer_b_input = gr.Textbox(label="Option B", placeholder="Answer B", lines=1, max_lines=1)
+        with gr.Row():
+            answer_c_input = gr.Textbox(label="Option C", placeholder="Answer C", lines=1, max_lines=1)
+            answer_d_input = gr.Textbox(label="Option D", placeholder="Answer D", lines=1, max_lines=1)
 
-    
-    get_answer_button = gr.Button("Get Answer", variant="primary")
-    final_answer_output = gr.Textbox(label="The Model's Answer", interactive=False)
+    # A larger, more prominent button
+    get_answer_button = gr.Button("✨ Predict Answer", variant="primary", size="lg")
+
+    # Distinct output box
+    final_answer_output = gr.Textbox(
+        label="Model Prediction",
+        interactive=False,
+        lines=2,
+        placeholder="The result will appear here..."
+    )
 
     get_answer_button.click(
         fn=get_answer,