启动web app问题

[gotothehill]

我启动web app的时候报错：

2024-06-27 15:25:16.467 Uncaught app exception Traceback (most recent call last): File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/torch/serialization.py", line 531, in _check_seekable f.seek(f.tell()) AttributeError: 'NoneType' object has no attribute 'seek'

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 600, in _run_script exec(code, module.dict) File "/AI/Streamer-Sales/app.py", line 29, in from utils.model_loader import RAG_RETRIEVER File "/AI/Streamer-Sales/utils/model_loader.py", line 18, in TTS_HANDLER = get_tts_model() File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/streamlit/runtime/caching/cache_utils.py", line 165, in wrapper return cached_func(*args, *kwargs) File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/streamlit/runtime/caching/cache_utils.py", line 194, in call return self._get_or_create_cached_value(args, kwargs) File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/streamlit/runtime/caching/cache_utils.py", line 221, in _get_or_create_cached_value return self._handle_cache_miss(cache, value_key, func_args, func_kwargs) File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/streamlit/runtime/caching/cache_utils.py", line 277, in _handle_cache_miss computed_value = self._info.func(func_args, **func_kwargs) File "/AI/Streamer-Sales/utils/tts/gpt_sovits/inference_gpt_sovits.py", line 503, in get_tts_model max_sec, t2s_model = change_gpt_weights(gpt_path, is_half) File "/AI/Streamer-Sales/utils/tts/gpt_sovits/inference_gpt_sovits.py", line 93, in change_gpt_weights dict_s1 = torch.load(gpt_path, map_location="cpu") File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/torch/serialization.py", line 986, in load with _open_file_like(f, 'rb') as opened_file: File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/torch/serialization.py", line 440, in _open_file_like return _open_buffer_reader(name_or_buffer) File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/torch/serialization.py", line 425, in init _check_seekable(buffer) File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/torch/serialization.py", line 534, in _check_seekable raise_err_msg(["seek", "tell"], e) File "/root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/torch/serialization.py", line 527, in raise_err_msg raise type(e)(msg) AttributeError: 'NoneType' object has no attribute 'seek'. You can only torch.load from a file that is seekable. Please pre-load the data into a buffer like io.BytesIO and try to load from it instead. /root/miniconda3/envs/streamer-sales/lib/python3.10/site-packages/streamlit/watcher/local_sources_watcher.py:196: UserWarning: Torchaudio's I/O functions now support par-call bakcend dispatch. Importing backend implementation directly is no longer guaranteed to work. Please use backend keyword with load/save/info function, instead of calling the udnerlying implementation directly. lambda m: [p for p in m.path._path],

请问，这个是什么问题呢？

[PeterH0323]

你好，需要确认下几个事情：

1、环境是否和我一致 2、请描述下您的启动步骤

[gotothehill]

你好，需要确认下几个事情：

1、环境是否和我一致 2、请描述下您的启动步骤

我是按照你的详细步骤来的

1. 环境搭建
2. 微调数据集准备：我直接使用的你提供的 train.jsonl，没有自己去微调
3. 训练：我下载的 internlm2-chat-7b-4bit模型训练的，使用的数据集为上面的train.jsonl，最终训练为 iter_2640.pth
4. 说明书生成：这步我跳过了
5. RAG 向量数据库：跳过了
6. 部署：我只是将命令中的 iter_340 都改为了我这边训练后的 iter_2640。4bit量化也做了，测试速度报错：需要给7b模型命名。最后启动，然后报错了。

[PeterH0323]

一般是使用 fp32 的模型作为预训练模型，训练好的模型再去量化 4bit ，您可以先使用我的模型，进行部署试下

[gotothehill]

一般是使用 fp32 的模型作为预训练模型，训练好的模型再去量化 4bit ，您可以先使用我的模型，进行部署试下

了解，也就是说训练时候，不能用4bit去训练是吧？ 直接使用你的模型，是把你的模型下载下来放到文件夹：/work_dirs/internlm2_chat_7b_qlora_custom_data/ 就可以直接使用吗？还是要其他配置

[PeterH0323]

对的，一般不建议用 4bit 去训练

不需要手动下载，直接启动就行，会自动下载的

[gotothehill]

对的，一般不建议用 4bit 去训练

不需要手动下载，直接启动就行，会自动下载的

我把云服务器重置，直接按照 快速体验 来启动，如下： git clone https://github.com/PeterH0323/Streamer-Sales.git cd Streamer-Sales conda env create -f environment.yml conda activate streamer-sales pip install -r requirements.txt

export USING_4BIT=true # 设置使用 4bit 模型 export KV_CACHE=0.05

streamlit run app.py --server.address=0.0.0.0 --server.port 7860

服务器用的A10 24G显卡，显卡驱动及cuda都已安装。还是报这个错误：AttributeError: 'NoneType' object has no attribute 'seek' ...

[PeterH0323]

服务启动的时候有自动下载权重吗？估计是网络问题导致下载出错，模型权重没加载成功

[gotothehill]

服务启动的时候有自动下载权重吗？估计是网络问题导致下载出错，模型权重没加载成功

问题找到了，系统没有安装unzip，导致每次启动都下载一个zip文件，然后又解压不了，就启动失败了。

谢谢

[PeterH0323]

谢谢反馈