Update app.py

app.py

@@ -1,30 +1,51 @@
 # app.py
 
 import gradio as gr
+import torch
 import pandas as pd
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import torch.nn.functional as F
 
-# Load cleaned dataset
-df = pd.read_csv("/tmp/eemm_cleaned.csv")
+# STEP 1: Define label names and load model
+label_names = [
+    "Cognition", "Affect", "Self", "Motivation", "Attention", "OB", "Context",
+    "Social", "Physical", "Psych"
+]
 
-# Dummy classifier function — replace with your model later
+model_path = "./results"
+tokenizer = AutoTokenizer.from_pretrained(model_path)
+model = AutoModelForSequenceClassification.from_pretrained(model_path)
+model.eval()
+
+# STEP 2: Prediction function
 def classify_question(text):
-    text = text.lower()
-    if "should" in text:
-        return "Cognition"
-    elif "feel" in text:
-        return "Affect"
-    else:
-        return "Unknown / Other"
+    with torch.no_grad():
+        inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
+        outputs = model(**inputs)
+        probs = torch.sigmoid(outputs.logits)[0]
+        predicted_labels = [label_names[i] for i, p in enumerate(probs) if p > 0.5]
+        if not predicted_labels:
+            return "No domain confidently predicted."
+        return ", ".join(predicted_labels)
+
+# STEP 3: Load example data if available
+try:
+    df = pd.read_csv("/tmp/eemm_cleaned.csv")
+    examples = df["clean_question"].dropna().head(5).tolist()
+except:
+    examples = ["I must be perfect to be accepted.", "I enjoy helping others."]
 
-# Gradio app
+# STEP 4: Gradio Interface
 demo = gr.Interface(
     fn=classify_question,
     inputs=gr.Textbox(label="Enter a question"),
-    outputs=gr.Textbox(label="Predicted category"),
-    examples=df["clean_question"].head().tolist(),
-    title="EEMM Exemplar Classifier",
-    description="This demo classifies survey questions into psychological domains. Replace this with your trained model for production.",
-    allow_flagging="never"  # ✅ Disable saving flagged examples
+    outputs=gr.Textbox(label="Predicted domains"),
+    examples=examples,
+    title="EEMM Multi-Label Classifier",
+    description="Classifies questions into multiple psychological domains.",
+    allow_flagging="never"
 )
 
-demo.launch(share=True, server_name="0.0.0.0")
+# STEP 5: Launch
+if __name__ == "__main__":
+    demo.launch(share=True, server_name="0.0.0.0")