Create app.py

app.py

@@ -0,0 +1,47 @@
+import gradio as gr
+from huggingface_hub import hf_hub_download
+from llama_cpp import Llama
+import os
+
+# Download the model (this will happen automatically when the space starts)
+model_path = hf_hub_download(
+    repo_id="bartowski/Gemma-2-2B-Instruct-GGUF",
+    filename="Gemma-2-2B-Instruct-Q4_K_M.gguf"
+)
+
+# Load the model
+llm = Llama(
+    model_path=model_path,
+    n_ctx=2048,      # Context length
+    n_threads=2,     # Use both CPUs
+    n_gpu_layers=0,  # CPU only
+    verbose=False
+)
+
+def chat_response(message, history):
+    # Format the prompt with conversation history
+    prompt = f"""<start_of_turn>user
+{message}<end_of_turn>
+<start_of_turn>model"""
+    
+    # Generate response
+    output = llm(
+        prompt,
+        max_tokens=256,
+        stop=["<end_of_turn>", "<start_of_turn>"],
+        temperature=0.7,
+        top_p=0.9
+    )
+    
+    return output['choices'][0]['text']
+
+# Create the chat interface
+demo = gr.ChatInterface(
+    fn=chat_response,
+    title="Hausa AI Assistant",
+    description="A simple AI assistant powered by Gemma 2"
+)
+
+# Launch the application
+if __name__ == "__main__":
+    demo.launch()