update

infer_intent.py

@@ -1,5 +1,4 @@
 from transformers import AutoTokenizer
-import torch
 import onnxruntime as ort
 import numpy as np
 import requests
@@ -36,7 +35,7 @@ class IntentClassifier:
 
     def find_intent(self, sequence, verbose=False):
         inputs = self.tokenizer(sequence,
-                return_tensors="pt",  # ONNX requires inputs in NumPy format
+                return_tensors="np",  # ONNX requires inputs in NumPy format
                 padding="max_length",  # Pad to max length
                 truncation=True,       # Truncate if the text is too long
                 max_length=64)