Update app.py

app.py

@@ -1,108 +1,37 @@
 import gradio as gr
-import time
-import random
-import matplotlib.pyplot as plt
 from transformers import pipeline
 
+# Choose a reliable model (you can change this)
 MODEL_NAME = "distilbert-base-uncased-finetuned-sst-2-english"
 
-# -------- Load the real inference model once at startup --------
-try:
-    sentiment_analyzer = pipeline("sentiment-analysis", model=MODEL_NAME)
-except Exception as e:
-    sentiment_analyzer = None
-    print("Model load failed:", e)
+# Load the model once
+sentiment_analyzer = pipeline("sentiment-analysis", model=MODEL_NAME)
 
-# -------- Simulated Training with Chart + Log --------
-def simulate_training(epochs, learning_rate):
-    import matplotlib.pyplot as plt
-    import random, time
-
-    accuracies = []
-    logs = []
-
-    # Start lower for visible climb
-    current_acc = 0.55 + random.uniform(0, 0.05)
-
-    for epoch in range(epochs):
-        time.sleep(0.3)
-
-        # Improvement scaled for realistic visible growth
-        improvement = learning_rate * random.uniform(20, 80)
-        current_acc += improvement
-
-        # Add small random noise and plateau effect
-        if epoch > epochs * 0.6:
-            current_acc += random.uniform(-0.015, 0.005)
-
-        # Clamp values between 0.55 and 0.95
-        current_acc = max(min(current_acc, 0.95), 0.55)
-        accuracies.append(round(current_acc, 3))
-        logs.append(f"Epoch {epoch+1}: Validation Accuracy = {current_acc:.3f}")
-
-    # Create the chart
-    plt.figure(figsize=(4, 2))
-    plt.plot(range(1, epochs + 1), accuracies, marker="o", color="tab:blue")
-    plt.axhline(y=accuracies[0], color="gray", linestyle="--", linewidth=1, label="Starting accuracy")
-    plt.title("Simulated Validation Accuracy per Epoch")
-    plt.xlabel("Epoch")
-    plt.ylabel("Accuracy")
-    plt.ylim(0.5, 1.0)
-    plt.grid(True)
-    plt.legend()
-
-    final_acc = round(accuracies[-1], 3)
-    return plt, "\n".join(logs), final_acc
-
-
-# -------- Real Inference Function --------
+# Function to analyze text
 def analyze_text(text):
-    if not sentiment_analyzer:
-        return "Model not loaded. Refresh the page and try again."
     try:
         result = sentiment_analyzer(text)[0]
         label = result.get("label", "UNKNOWN")
-        score = round(result.get("score", 0), 3)
+        score = round(result.get("score", 3))
         return f"{label} ({score})"
     except Exception as e:
         return f"Error during inference: {e}"
 
-# -------- Gradio Interface Layout --------
-with gr.Blocks(title="CIS1160 Training & Inference Demo") as demo:
-    gr.Markdown(
-        "### Part 1 – Simulated Training\n"
-        "Adjust the settings and click **Train** to visualize accuracy improvement.\n"
-        "_(This is a simulation – no real data is trained.)_"
-    )
-
-    with gr.Row():
-        epochs = gr.Slider(1, 10, value=3, step=1, label="Epochs")
-        lr = gr.Slider(0.001, 0.01, value=0.005, step=0.001, label="Learning Rate")
-
-    train_button = gr.Button("Train (Simulated)")
-    train_chart = gr.Plot(label="Simulated Validation Accuracy Chart")
-    train_log = gr.Textbox(label="Training Log", lines=8)
-    final_acc = gr.Number(label="Final Accuracy")
-
-    train_button.click(
-        simulate_training,
-        inputs=[epochs, lr],
-        outputs=[train_chart, train_log, final_acc]
-    )
-
+# Gradio interface
+with gr.Blocks(title="CIS1160 LLM Inference Demo") as demo:
     gr.Markdown(
-        "### Part 2 – Real Inference\n"
-        "Use the trained model to make predictions on new text.\n"
-        "_(This uses a real pre-trained model for sentiment analysis.)_"
+        """
+        ### Explore Inference
+        Enter any sentence and see how a trained model interprets it.
+        Try clearly positive, negative, and neutral examples.
+        """
     )
 
-    with gr.Row():
-        input_text = gr.Textbox(label="Enter text to analyze")
-        output_text = gr.Textbox(label="Sentiment Prediction")
+    input_text = gr.Textbox(label="Enter text to analyze", lines=2)
+    output_text = gr.Textbox(label="Model Prediction")
 
     run_button = gr.Button("Run Inference")
     run_button.click(analyze_text, inputs=input_text, outputs=output_text)
 
-# -------- Launch Application --------
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()