模型微调，加载训练数据后，报错： batch_size = input.size(0) AttributeError: 'str' object has no attribute 'size'

[feifei-love]

❓ Questions and Help

Before asking:

1. search the issues.
2. search the docs.

What is your question?

加载训练数据后，代码运行报错，

Code

[2024-02-26 16:35:10,943][root][INFO] - config.yaml is saved to: outputs/debug/ckpt/funasr2/exp2/config.yaml [2024-02-26 16:35:16,064][root][INFO] - init_param is not None: ('/root/.cache/modelscope/hub/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.pt',) [2024-02-26 16:35:16,065][root][INFO] - Loading pretrained params from /root/.cache/modelscope/hub/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.pt ckpt: /root/.cache/modelscope/hub/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.pt [2024-02-26 16:35:19,379][root][INFO] - total_num of samplers across ranks: 8635 [2024-02-26 16:35:19,379][root][INFO] - total_num of samplers across ranks: 2 [2024-02-26 16:35:19,380][root][WARNING] - distributed is not initialized, only single shard No checkpoint found at 'outputs/debug/ckpt/funasr2/exp2/model.pt', starting from scratch rank: 0, Training Epoch: 1: 0%| | 0/270 [00:00<?, ?it/s]Error executing job with overrides: ['+model=damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', '+model_revision=v2.0.4', '+train_data_set_list=/root/example_data/train/train.jsonl', '+valid_data_set_list=/root/example_data/train/val.jsonl', '++dataset_conf.batch_size=32', '++dataset_conf.batch_type=example', '++train_conf.max_epoch=2', '++dataset_conf.num_workers=4', '+output_dir=outputs/debug/ckpt/funasr2/exp2'] Traceback (most recent call last): File "/root/FunASR/funasr/bin/train.py", line 42, in main_hydra main(kwargs) File "/root/FunASR/funasr/bin/train.py", line 185, in main trainer.run() File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/funasr/train_utils/trainer.py", line 161, in run self._train_epoch(epoch) File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/funasr/train_utils/trainer.py", line 214, in _train_epoch for batch_idx, batch in enumerate(self.dataloader_train): File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 634, in next data = self._next_data() File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 1346, in _next_data return self._process_data(data) File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 1372, in _process_data data.reraise() File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/_utils.py", line 644, in reraise raise exception AttributeError: Caught AttributeError in DataLoader worker process 0. Original Traceback (most recent call last): File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/utils/data/_utils/worker.py", line 308, in _worker_loop data = fetcher.fetch(index) File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 51, in fetch data = [self.dataset[idx] for idx in possibly_batched_index] File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 51, in data = [self.dataset[idx] for idx in possibly_batched_index] File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/funasr/datasets/audio_datasets/datasets.py", line 61, in getitem speech, speech_lengths = extract_fbank(data_src, data_type=self.data_type, frontend=self.frontend, is_final=True) # speech: [b, T, d] File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/funasr/utils/load_utils.py", line 106, in extract_fbank data, data_len = frontend(data, data_len, kwargs) File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/root/.local/conda/envs/funasr/lib/python3.8/site-packages/funasr/frontends/wav_frontend.py", line 125, in forward batch_size = input.size(0) AttributeError: 'str' object has no attribute 'size'

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace. rank: 0, Training Epoch: 1: 0%|

What have you tried?

What's your environment?

• OS (e.g., Linux):
• FunASR Version (e.g., 1.0.10):
• ModelScope Version (e.g., 1.12.0):
• PyTorch Version (e.g., 2.0.0):
• How you installed funasr (pip, source):
• Python version:
• GPU (e.g., V100M32)
• CUDA/cuDNN version (e.g., cuda11.7):
• Docker version (无)
• Any other relevant information:

[gg22mm]

使用效果怎么样啊，跟 “讯飞听见” 比

[LauraGPT]

Bug has been fixed. Please ref to the docs: https://github.com/alibaba-damo-academy/FunASR/blob/main/docs/tutorial/README_zh.md

[juzstu]

I use funasr==1.0.23, still get the same error. File "/usr/local/lib/python3.8/dist-packages/funasr/frontends/wav_frontend.py", line 126, in forward batch_size = input.size(0) AttributeError: 'str' object has no attribute 'size'

[Lzzycola]

I use funasr==1.0.23, still get the same error. File "/usr/local/lib/python3.8/dist-packages/funasr/frontends/wav_frontend.py", line 126, in forward batch_size = input.size(0) AttributeError: 'str' object has no attribute 'size'

same with me

[xinliu9451]

我也遇到了这个报错，但是检查下来都是训练集音频地址对应的音频找不到，你可以先试着去读取数据集地址里面的每一个音频，并将捕获的异常打印出来

[farmer21cn]

这个问题解决了吗，是不是音频文件必须是http路径啊？

[xinliu9451]

这个问题解决了吗，是不是音频文件必须是http路径啊？

不是http也行，我当时是因为本地音频被破坏了，或者是对应地址的音频找不到。所以你可以试着去读取一下你的本地音频，然后排查一下是否有错

[farmer21cn]

谢谢，还真是你说的原因。

冯亚军 @.***

 

------------------ 原始邮件 ------------------ 发件人: "modelscope/FunASR" @.>; 发送时间: 2024年7月26日(星期五) 上午10:46 @.>; @.**@.>; 主题: Re: [modelscope/FunASR] 模型微调，加载训练数据后，报错： batch_size = input.size(0) AttributeError: 'str' object has no attribute 'size' (Issue #1389)

这个问题解决了吗，是不是音频文件必须是http路径啊？

不是http也行，我当时是因为本地音频被破坏了，或者是对应地址的音频找不到。所以你可以试着去读取一下你的本地音频，然后排查一下是否有错

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[segu-L]

确实，音频的json用的相对路径，换了个工作目录就报错了