update information

• deep_training

    10-24 initial asr ctc

install

• pip install -U -r requirements.txt
• 如果无法安装， 可以切换官方源 pip install -i https://pypi.org/simple -U -r requirements.txt

weigtht select one is suitable for you

支持且不限于以下权重

• wav2vec2-base-100h
• wav2vec2-base-960h
• wav2vec2-large-960h
• wav2vec2-large-960h-lv60-self
• wav2vec2-base
• wav2vec2-large
• wavlm-base-plus
• wavlm-base

data sample

• open_data https://github.com/ssbuild/open_data
• librispeech_asr_dummy https://huggingface.co/datasets/patrickvonplaten/librispeech_asr_dummy

单条数据示例

{"file": "../assets/librispeech_asr_dummy/1272-128104-0000.flac", "sentence": "MISTER QUILTER IS THE APOSTLE OF THE MIDDLE CLASSES AND WE ARE GLAD TO WELCOME HIS GOSPEL"}

infer

# infer_finetuning.py 推理微调模型
# infer_lora_finetuning.py 推理微调模型
 python infer_finetuning.py

training

    # 制作数据
    cd scripts
    bash train_full.sh -m dataset 
    or
    bash train_lora.sh -m dataset 
    
    注: num_process_worker 为多进程制作数据 ， 如果数据量较大 ， 适当调大至cpu数量
    dataHelper.make_dataset_with_args(data_args.train_file,mixed_data=False, shuffle=True,mode='train',num_process_worker=0)
    
    # 全参数训练 
        bash train_full.sh -m train
        
    # lora adalora ia3 
        bash train_lora.sh -m train

训练参数

训练参数

友情链接

• pytorch-task-example
• tf-task-example
• chatmoss_finetuning
• chatglm_finetuning
• t5_finetuning
• llm_finetuning
• llm_rlhf
• chatglm_rlhf
• t5_rlhf
• rwkv_finetuning
• baichuan_finetuning

纯粹而干净的代码

参考

https://github.com/facebookresearch/fairseq/tree/main/examples/wav2vec#wav2vec-20

Star History