From 9aa3a4d0226943e5b3d0fee54f697164ec90cfa2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E5=8C=97=E5=BF=B5?= Date: Tue, 31 Oct 2023 19:52:55 +0800 Subject: [PATCH] add paraformer-16k-en finetune pipeline --- .../finetune.py | 34 +++++++++++++++++++ 1 file changed, 34 insertions(+) create mode 100644 egs_modelscope/asr_vad_punc/speech_paraformer-large-vad-punc_asr_nat-en-16k-common-vocab10020/finetune.py diff --git a/egs_modelscope/asr_vad_punc/speech_paraformer-large-vad-punc_asr_nat-en-16k-common-vocab10020/finetune.py b/egs_modelscope/asr_vad_punc/speech_paraformer-large-vad-punc_asr_nat-en-16k-common-vocab10020/finetune.py new file mode 100644 index 000000000..7a33669e2 --- /dev/null +++ b/egs_modelscope/asr_vad_punc/speech_paraformer-large-vad-punc_asr_nat-en-16k-common-vocab10020/finetune.py @@ -0,0 +1,34 @@ +import os +from modelscope.metainfo import Trainers +from modelscope.trainers import build_trainer +from funasr.datasets.ms_dataset import MsDataset + + +def modelscope_finetune(params): + if not os.path.exists(params.output_dir): + os.makedirs(params.output_dir, exist_ok=True) + # dataset split ["train", "validation"] + ds_dict = MsDataset.load(params.data_path) + kwargs = dict( + model=params.model, + model_revision=params.model_revision, + data_dir=ds_dict, + dataset_type=params.dataset_type, + work_dir=params.output_dir, + batch_bins=params.batch_bins, + max_epoch=params.max_epoch, + lr=params.lr) + trainer = build_trainer(Trainers.speech_asr_trainer, default_args=kwargs) + trainer.train() + + +if __name__ == '__main__': + from funasr.utils.modelscope_param import modelscope_args + params = modelscope_args(model="damo/speech_paraformer-large-vad-punc_asr_nat-en-16k-common-vocab10020") + params.output_dir = "./checkpoint2" # m模型保存路径 + params.data_path = "./example_data/" # 数据路径 + params.dataset_type = "small" # 小数据量设置small，若数据量大于1000小时，请使用large + params.batch_bins = 1000 # batch size，如果dataset_type="small"，batch_bins单位为fbank特征帧数，如果dataset_type="large"，batch_bins单位为毫秒， + params.max_epoch = 50 # 最大训练轮数 + params.lr = 0.00005 # 设置学习率 + modelscope_finetune(params) \ No newline at end of file