Update handler.py

handler.py

@@ -34,7 +34,6 @@ class EndpointHandler():
             inputs = self.tokenizer(prompt, return_tensors="pt").to(self.device)
             input_size = inputs['input_ids'].size(1)
             input_ids = inputs["input_ids"].to(self.device)
-            inputs.pop("token_type_ids")  
             outputs = self.model(**inputs)
             last_token_logits = outputs.logits[:, -1, :]
             options_tokens_logits = last_token_logits[:, self.options_tokens].detach().cpu().numpy()