Update app.py

app.py

@@ -10,7 +10,7 @@ tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
 model = AutoModelForCausalLM.from_pretrained(
     MODEL_NAME,
     device_map="cpu",      # ✅ force CPU
-    torch_dtype="float32"  # ✅ safe on CPU
+    torch_dtype=torch.float16  # ✅ safe on CPU
 )
 
 # Build pipeline
@@ -36,7 +36,7 @@ def chatbot(message, history):
     # Generate response
     response = generator(
         context,
-        max_new_tokens=200,
+        max_new_tokens=128,
         do_sample=True,
         temperature=0.7,
         top_p=0.9