Update app.py

app.py

@@ -1,44 +1,41 @@
 import gradio as gr
-import pandas as pd
 import torch
-import joblib
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import joblib
 
-# ✅ Fixed: Use relative path (NOT absolute)
-model_repo = "campaign_bert_model/campaign_bert_model"
+# ✅ Set your model repo correctly — it should match your uploaded folder name in the Space
+model_repo = "./campaign-bert-model"  # DO NOT use './' if it's confusing HF. Use just 'campaign-bert-model' if needed
 
 # ✅ Load tokenizer and model
 tokenizer = AutoTokenizer.from_pretrained(model_repo, use_fast=False)
 model = AutoModelForSequenceClassification.from_pretrained(model_repo)
 
 # ✅ Load label encoder
-label_encoder = joblib.load("campaign_bert_model/label_encoder.pkl")
+label_encoder = joblib.load("label_encoder.pkl")
 
-# ✅ Prediction function
-def predict_template(client_interest, sentiment):
+# ✅ Inference function
+def predict(client_interest, sentiment):
     text = f"{client_interest} Sentiment: {sentiment}"
     inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
-    
     with torch.no_grad():
         outputs = model(**inputs)
-        prediction = torch.argmax(outputs.logits, dim=1).item()
-    
-    predicted_template = label_encoder.inverse_transform([prediction])[0]
-    
-    return f"🎯 Predicted Template: **{predicted_template}**"
+        probs = torch.nn.functional.softmax(outputs.logits, dim=1)
+        pred = torch.argmax(probs, dim=1).item()
+        label = label_encoder.inverse_transform([pred])[0]
+        confidence = probs[0][pred].item() * 100
+    return f"📩 Recommended Template: {label} (Confidence: {confidence:.2f}%)"
 
 # ✅ Gradio UI
-with gr.Blocks(title="Campaign Personalizer") as demo:
-    gr.Markdown("# 🧠 Campaign Personalizer\nPredict the best template based on client interest and sentiment.")
-    
-    with gr.Row():
-        client_input = gr.Textbox(label="Client Interest")
-        sentiment_input = gr.Textbox(label="Sentiment (e.g., Positive, Neutral, Negative)")
-
-    output = gr.Markdown()
-
-    predict_button = gr.Button("🔍 Predict Template")
-    predict_button.click(fn=predict_template, inputs=[client_input, sentiment_input], outputs=output)
-
-# ✅ Launch
-demo.launch()
+ui = gr.Interface(
+    fn=predict,
+    inputs=[
+        gr.Textbox(label="Client Interest"),
+        gr.Textbox(label="Sentiment")
+    ],
+    outputs=gr.Textbox(label="Predicted Template"),
+    title="🧠 Campaign Personalizer",
+    theme="soft",
+    description="Enter client interest and sentiment to get the best template suggestion."
+)
+
+ui.launch()