update handler

handler.py

@@ -28,9 +28,7 @@ class EndpointHandler():
         # self.pipeline = pipeline(task= "automatic-speech-recognition", model=model, tokenizer=tokenizer, feature_extractor=feature_extractor)
         self.pipeline.model.config.forced_decoder_ids = self.pipeline.tokenizer.get_decoder_prompt_ids(language="Chinese", task="transcribe")
         self.pipeline.model.generation_config.forced_decoder_ids = self.pipeline.model.config.forced_decoder_ids  # just to be sure!
-        # self.pipeline = pipeline(task= "automatic-speech-recognition", model=self.model)
-        # self.pipeline.model.config.forced_decoder_ids = self.pipeline.tokenizer.get_decoder_prompt_ids(language="Chinese", task="transcribe")
-        # self.pipeline.model.generation_config.forced_decoder_ids = self.pipeline.model.config.forced_decoder_ids
+        # self.pipeline = pipeline(task= "automatic-speech-recognition", model=self.model
 
     def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
         """
@@ -45,6 +43,4 @@ class EndpointHandler():
         # with torch.cuda.amp.autocast():
         with torch.no_grad():
             prediction = self.pipeline(inputs, generate_kwargs={"forced_decoder_ids": self.forced_decoder_ids}, max_new_tokens=255)
-            # prediction = self.pipeline(inputs, return_timestamps=False)
-            prediction['text'] = prediction['text'] + '????'
         return prediction