Update app.py

app.py

@@ -1,20 +1,41 @@
 import gradio as gr
 import torch
 import numpy as np
+import json
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
 from peft import PeftModel
+from huggingface_hub import hf_hub_download
+import os
 
 # --- 1. CONFIGURATION ---
-MODEL_METRICS = {
-    "Accuracy": "89.20%",
-    "F1_Score": "0.8931"
-}
-
 ADAPTER_REPO = "jvillar-sheff/ag-news-distilbert-lora"
 BASE_MODEL_ID = "distilbert-base-uncased"
 CLASS_NAMES = {0: "World", 1: "Sports", 2: "Business", 3: "Sci/Tech"}
 
-# --- 2. MODEL LOADING ---
+# --- 2. DYNAMIC METRICS LOADING ---
+def fetch_metrics():
+    """Downloads evaluation_report.json from the Model Hub."""
+    try:
+        file_path = hf_hub_download(repo_id=ADAPTER_REPO, filename="evaluation_report.json")
+        with open(file_path, "r") as f:
+            data = json.load(f)
+        
+        # Extract numbers
+        acc = data['overall_metrics']['Accuracy']
+        f1 = data['overall_metrics']['F1 Macro']
+        
+        return {
+            "Accuracy": f"{acc:.2%}",
+            "F1_Score": f"{f1:.4f}"
+        }
+    except Exception as e:
+        print(f"Error loading metrics: {e}")
+        return {"Accuracy": "N/A", "F1_Score": "N/A"}
+
+# Load metrics on app startup
+MODEL_METRICS = fetch_metrics()
+
+# --- 3. MODEL LOADING ---
 def load_model():
     print("Loading Base Model...")
     base_model = AutoModelForSequenceClassification.from_pretrained(
@@ -30,6 +51,7 @@ def load_model():
     print("Loading Adapters...")
     model = PeftModel.from_pretrained(base_model, ADAPTER_REPO)
     
+    # Force CPU for Free Tier Spaces
     device = torch.device("cpu")
     model.to(device)
     model.eval()
@@ -37,7 +59,7 @@ def load_model():
 
 model, tokenizer, device = load_model()
 
-# --- 3. PREDICTION LOGIC ---
+# --- 4. PREDICTION LOGIC ---
 def predict(text):
     if not text.strip():
         return None, None, None
@@ -60,35 +82,36 @@ def predict(text):
     # 2. Create Probability Dict for the Chart
     class_probs = {CLASS_NAMES[i]: float(probs[i]) for i in range(len(CLASS_NAMES))}
 
-    # 3. Create HTML for the "Confidence Badge" (Mimicking Streamlit)
+    # 3. Create HTML for the "Confidence Badge"
     if conf > 0.85:
-        bg_color, txt_color = "#d4edda", "#155724" # Green
+        bg_color, txt_color, icon = "#d4edda", "#155724", "↑" # Green
     elif conf > 0.60:
-        bg_color, txt_color = "#fff3cd", "#856404" # Yellow
+        bg_color, txt_color, icon = "#fff3cd", "#856404", "~" # Yellow
     else:
-        bg_color, txt_color = "#f8d7da", "#721c24" # Red
+        bg_color, txt_color, icon = "#f8d7da", "#721c24", "↓" # Red
         
     badge_html = f"""
     <div style='background-color: {bg_color}; color: {txt_color}; 
-    padding: 8px 12px; border-radius: 5px; display: inline-block; font-weight: bold; font-size: 16px;'>
-    Confidence: {conf:.2%}
+    padding: 8px 16px; border-radius: 5px; display: inline-block; font-weight: bold; font-size: 16px;'>
+    {icon} Confidence: {conf:.2%}
     </div>
     """
     
     # Return: Label Text, Badge HTML, Chart Data
     return f"# {pred_label}", badge_html, class_probs
 
-# --- 4. UI LAYOUT (gr.Blocks) ---
-with gr.Blocks() as demo:
+# --- 5. UI LAYOUT (gr.Blocks) ---
+# Using Soft theme (requires newer Gradio version in requirements.txt)
+# If it fails, remove theme=gr.themes.Soft()
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
     
-    # Title
     gr.Markdown("# 📰 NLP News Classifier")
     gr.Markdown("Classify news articles into World, Sports, Business, or Sci/Tech using DistilBERT + LoRA.")
 
     # -- The "Green Banner" (HTML) --
     gr.HTML(f"""
     <div style="background-color: #d1e7dd; color: #0f5132; padding: 15px; border-radius: 5px; border: 1px solid #badbcc; margin-bottom: 20px;">
-        ✅ <b>Model Performance:</b> Accuracy: {MODEL_METRICS['Accuracy']} | F1 Score: {MODEL_METRICS['F1_Score']}
+        ✅ <b>Model Performance (Test Set):</b> Accuracy: {MODEL_METRICS['Accuracy']} | F1 Score: {MODEL_METRICS['F1_Score']}
     </div>
     """)
 
@@ -121,7 +144,7 @@ with gr.Blocks() as demo:
             out_badge = gr.HTML()
             
             gr.Markdown("### Probability Breakdown")
-            # Output 3: Bar Chart (Label component handles this beautifully)
+            # Output 3: Bar Chart
             out_chart = gr.Label(num_top_classes=4, label="Confidence Scores")
 
     # Wire up the button