Update app.py

app.py

@@ -4,27 +4,31 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 from peft import PeftModel, PeftConfig
 import spaces
 
+# Check if CUDA is available and set the device
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+print(f"Using device: {device}")
+
 # Load model and tokenizer
 MODEL_PATH = "sagar007/phi2_finetune"
-
 tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH, trust_remote_code=True)
 tokenizer.pad_token = tokenizer.eos_token
 
 base_model = AutoModelForCausalLM.from_pretrained(
     "microsoft/phi-2",
-    torch_dtype=torch.float32,  # Use float32 for CPU
+    torch_dtype=torch.float16 if device.type == "cuda" else torch.float32,
     device_map="auto",
     trust_remote_code=True
 )
 
 peft_config = PeftConfig.from_pretrained(MODEL_PATH)
 model = PeftModel.from_pretrained(base_model, MODEL_PATH)
+model.to(device)
 model.eval()
 
 @spaces.GPU(duration=60)
 def generate_response(instruction, max_length=512):
     prompt = f"Instruction: {instruction}\nResponse:"
-    inputs = tokenizer(prompt, return_tensors="pt")
+    inputs = tokenizer(prompt, return_tensors="pt").to(device)
     
     with torch.no_grad():
         outputs = model.generate(