Update app.py

app.py

@@ -1,58 +1,43 @@
 import gradio as gr
 from transformers import pipeline
 import pandas as pd
-import io
 
-# Load once
-classifier = pipeline("zero-shot-classification", model="facebook/bart-large-mnli")
+MODEL_MAP = {
+    "MoritzLaurer/deberta-v3-large-zeroshot-v2.0": "MoritzLaurer/deberta-v3-large-zeroshot-v2.0",
+    "MoritzLaurer/mDeBERTa-v3-base-xnli-multilingual-nli-2mil7": "MoritzLaurer/mDeBERTa-v3-base-xnli-multilingual-nli-2mil7",
+    "joeddav/xlm-roberta-large-xnli": "joeddav/xlm-roberta-large-xnli"
+}
 
-LIKERT_OPTIONS = ["Strongly disagree", "Disagree", "Neutral", "Agree", "Strongly agree"]
+def classify_items(model_name, items_text, labels_text):
+    classifier = pipeline("zero-shot-classification", model=MODEL_MAP[model_name])
+    items = [item.strip() for item in items_text.split("\n") if item.strip()]
+    labels = [label.strip() for label in labels_text.split(",") if label.strip()]
 
-response_table = []
+    results = []
+    for item in items:
+        out = classifier(item, labels, multi_label=True)
+        scores = {label: prob for label, prob in zip(out["labels"], out["scores"])}
+        scores["item"] = item
+        results.append(scores)
+
+    df = pd.DataFrame(results).fillna(0)
+    return df, df.to_csv(index=False)
 
-def classify_likert(questions_text):
-    questions = [q.strip() for q in questions_text.strip().split("\n") if q.strip()]
-    
-    global response_table
-    response_table = []
-    output_lines = []
-
-    for i, question in enumerate(questions, 1):
-        result = classifier(question, LIKERT_OPTIONS, multi_label=False)
-        probs = dict(zip(result['labels'], result['scores']))
-        output_lines.append(f"{i}. {question}")
-        for label in LIKERT_OPTIONS:
-            prob = round(probs.get(label, 0.0), 3)
-            output_lines.append(f"→ {label}: {prob}")
-        output_lines.append("")
-
-        row = {"Item #": i, "Item": question}
-        row.update({label: round(probs.get(label, 0.0), 3) for label in LIKERT_OPTIONS})
-        response_table.append(row)
-
-    return "\n".join(output_lines)
-
-def download_csv():
-    global response_table
-    if not response_table:
-        return None
-    df = pd.DataFrame(response_table)
-    csv_buffer = io.StringIO()
-    df.to_csv(csv_buffer, index=False)
-    return csv_buffer.getvalue()
-
-# Gradio interface
 with gr.Blocks() as demo:
-    gr.Markdown("# Likert-Style Zero-Shot Classifier")
-    gr.Markdown("Paste questionnaire items. Each will be classified into: Strongly disagree → Strongly agree, with probabilities.")
+    gr.Markdown("### Zero-Shot Questionnaire Classifier")
+
+    with gr.Row():
+        model_choice = gr.Dropdown(choices=list(MODEL_MAP.keys()), label="Choose a model")
+    
+    item_input = gr.Textbox(label="Enter questionnaire items (one per line)", lines=5, placeholder="e.g., I enjoy social gatherings.\nI prefer planning over spontaneity.")
+    label_input = gr.Textbox(label="Enter response options (comma-separated)", placeholder="e.g., Strongly disagree, Disagree, Neutral, Agree, Strongly agree")
 
-    questions_input = gr.Textbox(label="Enter multiple items (one per line)", lines=10, placeholder="e.g.\nI feel in control of my life.\nI enjoy being around others...")
-    output_box = gr.Textbox(label="Classification Output", lines=20)
-    submit_btn = gr.Button("Classify Items")
-    csv_btn = gr.Button("📥 Download CSV")
-    file_output = gr.File(label="Download CSV", visible=False)
+    run_button = gr.Button("Classify")
+    output_table = gr.Dataframe(label="Classification Results")
+    download_csv = gr.File(label="Download CSV")
 
-    submit_btn.click(fn=classify_likert, inputs=questions_input, outputs=output_box)
-    csv_btn.click(fn=download_csv, inputs=[], outputs=file_output)
+    run_button.click(fn=classify_items, 
+                     inputs=[model_choice, item_input, label_input], 
+                     outputs=[output_table, download_csv])
 
 demo.launch()