Merge pull request #1053 from alibaba-damo-academy/dev_lzr_en

Lizerui9926

2023-11-02 8c904ecadde91d19d97a011377cc978459d601ce

Merge pull request #1053 from alibaba-damo-academy/dev_lzr_en

support paraformer-16k-en finetune

 egs_modelscope/asr_vad_punc/speech_paraformer-large-vad-punc_asr_nat-en-16k-common-vocab10020/finetune.py

New file
@@ -0,0 +1,35 @@
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 = "./checkpoint"              # 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
    params.model_revision = "v1.0.1"
    modelscope_finetune(params)

 funasr/bin/build_trainer.py

@@ -548,6 +548,7 @@
    init_param = modelscope_dict['init_model']
    cmvn_file = modelscope_dict['cmvn_file']
    seg_dict_file = modelscope_dict['seg_dict']
    bpemodel = modelscope_dict['bpemodel']

    # overwrite parameters
    with open(config) as f:
@@ -581,6 +582,10 @@
        args.seg_dict_file = seg_dict_file
    else:
        args.seg_dict_file = None
    if os.path.exists(bpemodel):
        args.bpemodel = bpemodel
    else:
        args.bpemodel = None
    args.data_dir = data_dir
    args.train_set = train_set
    args.dev_set = dev_set