Update handler.py

handler.py

@@ -8,7 +8,8 @@ class EndpointHandler():
         # Preload all the elements you are going to need at inference.
         # pseudo:
         self.tokenizer = AutoTokenizer.from_pretrained("philschmid/falcon-40b-instruct-GPTQ-inference-endpoints", use_fast=False)
-        self.model = AutoGPTQForCausalLM.from_quantized("philschmid/falcon-40b-instruct-GPTQ-inference-endpoints", device="cuda:0", use_triton=False, use_safetensors=True, trust_remote_code=True)
+        self.model = AutoGPTQForCausalLM.from_quantized("philschmid/falcon-40b-instruct-GPTQ-inference-endpoints", device="cuda:0", use_triton=False, use_safetensors=True, torch_dtype=torch.float32, trust_remote_code=True)
+
 
 
     def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]: