Closed xuhongtian closed 11 months ago
按照你提供的https://alibaba-damo-academy.github.io/FunASR/en/modelscope_pipeline/quick_start_zh.html#id9,我在本地服务器进行模型下载和适配,还是依然存在服务器无法加载离线模型。 code: from modelscope.pipelines import pipeline from modelscope.utils.constant import Tasks
modelpath = "/data/modelscope_bak/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" asr_pipeline = pipeline(task=Tasks.auto_speech_recognition,model=modelpath) wav_name = "2023110200000949.wav" asr_result = asr_pipeline(audio_in=wav_name) print(asr_result) 模型存放本地地址如下:
config配置如下
发现依然存在模型启动后还是会默认去读取cache,从而无法首先读取本地模型。
os:python3.9的 modelscope=1.9.4 funasr=0.8.4,torch=2.0.1. inference_pipline = pipeline(task=Tasks.auto_speech_recognition,model="/data/asr_model/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch") 本地configuration.json配置 { "framework": "pytorch", "task" : "auto-speech-recognition", "model" : { "type" : "generic-asr", "am_model_name" : "model.pb", "model_config" : { "type": "pytorch", "code_base": "funasr", "mode": "paraformer", "lang": "zh-cn", "batch_size": 1, "am_model_config": "config.yaml", "asr_model_config": "decoding.yaml", "mvn_file": "am.mvn", "model":"/data/asr_model/peech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch", "lm_model": "/data/asr_model/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch" } }, "pipeline": { "type":"asr-inference" } } 我的配置文件应该均已下载到本地,列表如下 请问下这是什么问题?