Update app.py

app.py

@@ -5,18 +5,17 @@ import spaces
 
 model_id = "microsoft/phi-2"
 
-# 🌟 Auto-detect device (CPU/GPU)
-device = "cuda"
-precision = torch.float16
+# Auto-detect device (but since your space is ZeroGPU, this should default to CPU)
+device = "cuda" if torch.cuda.is_available() else "cpu"
+precision = torch.float16 if device == "cuda" else torch.float32
 
-# Load model and tokenizer (CPU + float32)
+# Load model and tokenizer
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(
     model_id,
     torch_dtype=precision
 ).to(device)
 
-
 # Chat function
 @spaces.GPU
 def chat_with_bot(message, history):
@@ -38,10 +37,11 @@ def chat_with_bot(message, history):
 
     decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
     response = decoded[len(prompt):].strip().split("\n")[0]
+    # IMPORTANT: Append message pair as a list rather than a tuple.
     history.append([message, response])
     return response, history
 
-# ✅ gr.ChatInterface expects fn(message, history) → (response, updated_history)
+# Gradio ChatInterface expects fn(message, history) → (response, updated_history)
 gr.ChatInterface(
     fn=chat_with_bot,
     title="Phi-2 Chatbot (CPU-friendly)",