TypeError: _sanitize_parameters() got an unexpected keyword argument 'low_cpu_mem_usage'

[37llll]

Describe the Question

Please provide a clear and concise description of what the que---------------------------------------------------------------------------

运行例子时遇到了这个错误，应该如何解决

TypeError Traceback (most recent call last)
Cell In[1], line 5
1 from parrots import SpeechRecognition
4 if name == 'main':
----> 5 m = SpeechRecognition('/app/pretrained_models/Belle-distilwhisper-large-v2-zh', low_cpu_mem_usage=False)
6 r = m.recognize_speech_from_file('./output.wav')
7 print('[提示] 语音识别结果：', r)

File /app/project/ASR-TTS/parrots/asr.py:81, in SpeechRecognition.init(self, model_name_or_path, use_cuda, cuda_device, max_new_tokens, chunk_length_s, batch_size, torch_dtype, use_flash_attention_2, language, **kwargs)
78 self.model.to(self.device)
80 self.processor = AutoProcessor.from_pretrained(model_name_or_path)
---> 81 self.pipe = pipeline(
82 "automatic-speech-recognition",
83 model=self.model,
84 tokenizer=self.processor.tokenizer,
85 feature_extractor=self.processor.feature_extractor,
86 device=self.device,
87 torch_dtype=torch_dtype,
88 max_new_tokens=max_new_tokens,
89 batch_size=batch_size,
90 chunk_length_s=chunk_length_s,
91 **kwargs
92 )
93 if language == 'zh':
94 self.pipe.model.config.forced_decoder_ids = (
95 self.pipe.tokenizer.get_decoder_prompt_ids(
96 language=language,
97 task="transcribe"
98 )
99 )

File ~/miniconda3/envs/speech_recognition/lib/python3.9/site-packages/transformers/pipelines/init.py:1108, in pipeline(task, model, config, tokenizer, feature_extractor, image_processor, framework, revision, use_fast, token, device, device_map, torch_dtype, trust_remote_code, model_kwargs, pipeline_class, **kwargs)
1105 if device is not None:
1106 kwargs["device"] = device
-> 1108 return pipeline_class(model=model, framework=framework, task=task, **kwargs)

File ~/miniconda3/envs/speech_recognition/lib/python3.9/site-packages/transformers/pipelines/automatic_speech_recognition.py:220, in AutomaticSpeechRecognitionPipeline.init(self, model, feature_extractor, tokenizer, decoder, device, torch_dtype, **kwargs)
217 else:
218 self.type = "ctc"
--> 220 super().init(model, tokenizer, feature_extractor, device=device, torch_dtype=torch_dtype, **kwargs)

File ~/miniconda3/envs/speech_recognition/lib/python3.9/site-packages/transformers/pipelines/base.py:894, in Pipeline.init(self, model, tokenizer, feature_extractor, image_processor, modelcard, framework, task, args_parser, device, torch_dtype, binary_output, **kwargs)
892 self._batch_size = kwargs.pop("batch_size", None)
893 self._num_workers = kwargs.pop("num_workers", None)
--> 894 self._preprocess_params, self._forward_params, self._postprocess_params = self._sanitize_parameters(**kwargs)
896 # Pipelines calling generate: if the tokenizer has a pad token but the model doesn't, set it in the
897 # forward params so that generate is aware of the pad token.
898 if (
899 self.tokenizer is not None
900 and self.model.can_generate()
901 and self.tokenizer.pad_token_id is not None
902 and self.model.generation_config.pad_token_id is None
903 ):

TypeError: _sanitize_parameters() got an unexpected keyword argument 'low_cpu_mem_usage'stion is.

[shibing624]

transformers啥版本？

[37llll]

transformers啥版本？
感谢回复
我的是transformers 4.40.2
请问需要的是什么版本

[shibing624]

1。最新版本的transformers是可以的；
2.保证gpu的显存足够，一般大于8GB；
3.low_cpu_mem_usage这个参数删掉。