Update app.py

app.py

@@ -6,8 +6,10 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
 model_name = "Tap-M/Luna-AI-Llama2-Uncensored"
 
 bnb_config = BitsAndBytesConfig(
-    load_in_4bit=True,  # Enable 4-bit quantization
-    bnb_4bit_compute_dtype=torch.float16
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",       # Must be 'nf4' for CPU compatibility
+    bnb_4bit_compute_dtype="float32", # Use float32 for CPU computation
+    bnb_4bit_use_double_quant=True
 )
 
 model = AutoModelForCausalLM.from_pretrained(
@@ -15,7 +17,6 @@ model = AutoModelForCausalLM.from_pretrained(
     device_map="auto",            # Auto-distribute across GPU/CPU
     quantization_config=bnb_config,
     offload_folder="./offload",  # Temporary directory
-    low_cpu_mem_usage=True,      # Reduces CPU memory spikes
     trust_remote_code=True        # Required for some models
 )