Create app.py

app.py

@@ -0,0 +1,58 @@
+import gradio as gr
+from union.remote import UnionRemote
+from flytekit.types.file import FlyteFile
+
+remote = UnionRemote()
+
+# Fetch recent executions
+recent_executions = remote.recent_executions(limit=10)
+executions = [
+    e for e in recent_executions if e.spec.launch_plan.name == "bert-fine-tune.bert_ft"
+]
+recent_ex_id = executions[0].id.name
+execution = remote.fetch_execution(name=recent_ex_id)
+
+ft_llm = execution.outputs["o0"].remote_source
+model_cache = execution.outputs["o1"].remote_source
+
+predict_task = remote.fetch_task(name="bert-fine-tune.predict_sentiment")
+
+
+def execute_flyte_task(text):
+    inputs = {
+        "text": text,
+        "model_cache_dir": model_cache,
+        "model": FlyteFile(ft_llm)
+    }
+    execution = remote.execute(predict_task, inputs=inputs, wait=True)
+
+    # Extract the response
+    response = execution.outputs["o0"]
+
+    # Format the response
+    sentiment = response['label']
+    score = response['score']
+
+    # Determine color based on sentiment
+    color = "red" if sentiment == "NEGATIVE" else "green"
+
+    # Format the output HTML
+    output_html = f"""
+    <div style="text-align: center;">
+        <h2>Sentiment: <span style="color: {color};">{sentiment}</span></h2>
+        <p>Confidence Score: {score:.2f}</p>
+    </div>
+    """
+
+    return output_html
+
+
+# Launch Gradio app
+iface = gr.Interface(
+    fn=execute_flyte_task,
+    inputs=["text"],
+    outputs=gr.HTML(),  # Change output to HTML for better formatting
+    live=False,
+)
+
+iface.launch(debug=True)