search

modelscope / FunASR

A Fundamental End-to-End Speech Recognition Toolkit and Open Source SOTA Pretrained Models, Supporting Speech Recognition, Voice Activity Detection, Text Post-processing etc.
https://www.funasr.com
Other
6.18k stars 659 forks source link

How to decode Lora finetune model #934

Open lhanzl opened 1 year ago

lhanzl commented 1 year ago

ubuntu20.04, funasr&modelscope is last version. Lora finetune model can not decode:

RuntimeError：AutomaticSpeechRecognitionPipeline：Error（s）in loading state_dict for Paraformer： unexpected key(s) in state_dict: "encoder.encoders0.0.self_attn.linear_q_k_v.lora_A"

hnluo commented 1 year ago

Please list your envs:

Package Version：funasr version（pip list） Model： Command： Details： Error log：

lhanzl commented 1 year ago

Package Version：0.7.6 Model：Lora finetuned speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch model Command：python infer.py Details：when inference lora finetuned speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch model Error log： RuntimeError：AutomaticSpeechRecognitionPipeline：Error（s）in loading state_dict for Paraformer： unexpected key(s) in state_dict: "encoder.encoders0.0.self_attn.linear_q_k_v.lora_A"

LGoldenDragon commented 7 months ago

ubuntu20.04, funasr&modelscope is last version. Lora finetune model can not decode:

RuntimeError：AutomaticSpeechRecognitionPipeline：Error（s）in loading state_dict for Paraformer： unexpected key(s) in state_dict: "encoder.encoders0.0.self_attn.linear_q_k_v.lora_A"

请问一下这个问题解决了吗？我也出现了这个问题