Create app.py

app.py

@@ -0,0 +1,17 @@
+import gradio as gr
+from llama_cpp import Llama
+
+llm = Llama(
+    model_path="gemma-4-E4B-it-Q4_K_M.gguf",  # polku malliin
+    n_ctx=8192,
+    n_threads=2,  # sun 2 vCPU
+    n_gpu_layers=0  # CPU only
+)
+
+def chat(message, history):
+    response = llm.create_chat_completion(
+        messages=[{"role": "user", "content": message}]
+    )
+    return response['choices'][0]['message']['content']
+
+gr.ChatInterface(chat).launch()