Create app.py

app.py

@@ -0,0 +1,22 @@
+import gradio as gr
+from huggingface_hub import InferenceClient
+
+client = InferenceClient(model="http://127.0.0.1:8080")
+
+def inference(message, history):
+    partial_message = ""
+    for token in client.text_generation(message, max_new_tokens=20, stream=True):
+        partial_message += token
+        yield partial_message
+
+gr.ChatInterface(
+    inference,
+    chatbot=gr.Chatbot(height=300),
+    textbox=gr.Textbox(placeholder="Chat with me!", container=False, scale=7),
+    description="This is the demo for Gradio UI consuming TGI endpoint with LLaMA 7B-Chat model.",
+    title="Gradio 🤝 TGI",
+    examples=["Are tomatoes vegetables?"],
+    retry_btn="Retry",
+    undo_btn="Undo",
+    clear_btn="Clear",
+).queue().launch()