wealthcoders commited on
Commit
1fe38b4
·
verified ·
1 Parent(s): 5bdd01a

Update handler.py

Browse files
Files changed (1) hide show
  1. handler.py +6 -1
handler.py CHANGED
@@ -52,5 +52,10 @@ class EndpointHandler:
52
  inputs = inputs.to(self.model.device)
53
 
54
  generated_ids = self.model.generate(**inputs, max_new_tokens=2048)
55
- output_text = self.processor.batch_decode(generated_ids, skip_special_tokens=True)
 
 
 
 
 
56
  return output_text[0]
 
52
  inputs = inputs.to(self.model.device)
53
 
54
  generated_ids = self.model.generate(**inputs, max_new_tokens=2048)
55
+ generated_ids_trimmed = [
56
+ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
57
+ ]
58
+ output_text = processor.batch_decode(
59
+ generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
60
+ )
61
  return output_text[0]