add app.py

app.py

@@ -0,0 +1,38 @@
+from transformers import pipeline
+from transformers import AutoModelForSeq2SeqLM
+from transformers import AutoTokenizer
+import gradio as gr
+import torch
+
+# Load trained model
+model = AutoModelForSeq2SeqLM.from_pretrained("chkla/parlbert-topic-german")
+tokenizer = AutoTokenizer.from_pretrained("bert-base-german-cased")
+pipeline_classification_topics = pipeline("text-classification", model="chkla/parlbert-topic-german", tokenizer="bert-base-german-cased", return_all_scores=False)
+
+def predict_topic(input_text):
+    reframed_text = pipeline_classification_topics(input_text)
+    return reframed_text
+
+# Build Gradio interface
+with gr.Blocks() as demo:
+    # Instruction
+    gr.Markdown('''Topic Modelling for German Political Texts''')
+
+    # Input text to be reframed
+    text = gr.Textbox(label="Text")
+
+    # Trigger button for topic prediction
+    greet_btn = gr.Button("Predict Topic")
+    greet_btn.click(fn=predict_topic, inputs=[text], outputs=[gr.outputs.Textbox(label="Topic")])
+
+    # Default examples of text and strategy pairs for user to have a quick start
+    gr.Markdown("## Examples")
+    # gr.Examples(
+    #     [["I have a lot of homework to do today.", "self_affirmation"], ["So stressed about the midterm next week.", "optimism"], ["I failed my math quiz I am such a loser.", "growth"]], 
+    #     [text], best_output, show_reframe_change, cache_examples=False, run_on_click=False
+    # )
+
+    # Link to paper and Github repo
+    gr.Markdown('''For more details: You can read our [paper]() or access our [code]().''')
+
+demo.launch()