[bug]: 服务端启动报错

UlinoyaPed commented 4 months ago

Details

Process Process-2: Traceback (most recent call last): File "multiprocessing\process.py", line 315, in _bootstrap File "multiprocessing\process.py", line 108, in run File "D:\Tools\CapsWriter\CapsWriter-Offline-Windows-64bit\util\server_init_recognizer.py", line 44, in init_recognizer punc_model = CT_Transformer(ModelPaths.punc_model_dir, quantize=True) File "funasr_onnx\punc_bin.py", line 69, in __init__ File "funasr_onnx\utils\utils.py", line 209, in __init__ File "onnxruntime\capi\onnxruntime_inference_collection.py", line 419, in __init__ File "onnxruntime\capi\onnxruntime_inference_collection.py", line 452, in _create_inference_session onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Load model from models\punc_ct-transformer_cn-en\model_quant.onnx failed:bad allocation

描述Bug Describe the bug

服务端启动时报错，报错信息如下

Process Process-2:
Traceback (most recent call last):
  File "multiprocessing\process.py", line 315, in _bootstrap
  File "multiprocessing\process.py", line 108, in run
  File "D:\Tools\CapsWriter\CapsWriter-Offline-Windows-64bit\util\server_init_recognizer.py", line 44, in init_recognizer
    punc_model = CT_Transformer(ModelPaths.punc_model_dir, quantize=True)
  File "funasr_onnx\punc_bin.py", line 69, in __init__
  File "funasr_onnx\utils\utils.py", line 209, in __init__
  File "onnxruntime\capi\onnxruntime_inference_collection.py", line 419, in __init__
  File "onnxruntime\capi\onnxruntime_inference_collection.py", line 452, in _create_inference_session
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Load model from models\punc_ct-transformer_cn-en\model_quant.onnx failed:bad allocation

如何再现 To Reproduce

再现Bug的步骤：

models已下载并复制到models文件夹
双击start_server.exe启动服务端，模块成功加载，语音模块成功加载
标点模型等待较长时间后加载失败并显示报错，多次尝试结果相同

截屏 Screenshots

/

出现问题的设备 Devices with issues

设备：Win10 20H2
软件&版本：v1.0

备注 Additional context

models/模型测试文件夹中的python文件也不能正常运行

D:\Tools\CapsWriter\CapsWriter-Offline-Windows-64bit\models\模型测试>python D:\Tools\CapsWriter\CapsWriter-Offline-Windows-64bit\models\模型测试\03-01-标点模型测试.py
Traceback (most recent call last):
  File "D:\Tools\CapsWriter\CapsWriter-Offline-Windows-64bit\models\模型测试\03-01-标点模型测试.py", line 1, in <module>
    from funasr_onnx import CT_Transformer
ModuleNotFoundError: No module named 'funasr_onnx'

gongqf commented 4 months ago

03-01-标点模型测试.py 要复制到模型文件夹下才能运行

HaujetZhao commented 4 months ago

bad allocation ，有可能是内存不足导致。

你的电脑内存多大，设置的虚拟内存多大？

UlinoyaPed commented 4 months ago

bad allocation ，有可能是内存不足导致。

你的电脑内存多大，设置的虚拟内存多大？

🤔 可能是这个原因，我电脑内存只有4G，虚拟内存也不太大。

不过mem reduct在过程中全程开启，系统空载时其内存占用显示为40%, 整个运行过程中显示的内存占用一直未超过60%. 如果是内存不够为什么不会把内存占满？

HaujetZhao commented 4 months ago

虚拟内存放大到 16G

qsxmj commented 2 months ago

Process Process-2: Traceback (most recent call last): File "multiprocessing\process.py", line 315, in _bootstrap File "multiprocessing\process.py", line 108, in run File "D:\Ai\CapsWriter-Offline-Windows-64bit-withmodles\util\server_init_recognizer.py", line 44, in init_recognizer punc_model = CT_Transformer(ModelPaths.punc_model_dir, quantize=True) File "funasr_onnx\punc_bin.py", line 83, in init File "funasr_onnx\utils\utils.py", line 302, in code_mix_split_words_jieba File "jieba__init.py", line 416, in load_userdict File "jieba__init__.py", line 427, in add_word File "jieba\init__.py", line 464, in suggest_freq ZeroDivisionError: float division by zero 这是我遇到的报错与你复现经历的相似

HaujetZhao / CapsWriter-Offline