support GPU inference

modeling_hhem_v2.py

@@ -59,7 +59,7 @@ class HHEMv2ForSequenceClassification(PreTrainedModel):
         tokenizer = self.tokenzier
         pair_dict = [{'text1': pair[0], 'text2': pair[1]} for pair in text_pairs]
         inputs = tokenizer(
-            [self.prompt.format(**pair) for pair in pair_dict], return_tensors='pt', padding=True)
+            [self.prompt.format(**pair) for pair in pair_dict], return_tensors='pt', padding=True).to(self.t5.device)
         self.t5.eval()
         with torch.no_grad():
             outputs = self.t5(**inputs)