记录YOLOv5运行train.py出现的问题

目录

1.AttributeError: module 'numpy' has no attribute 'int'

2.RuntimeError: result type Float can't be cast to the desired output type long int

3.运行很慢很慢，gpu占用率低低低

4.OutOfMemoryError: CUDA out of memory. Tried to allocate 50.00 MiB (GPU 0; 23.65 GiB total capacity; 21.91 GiB already allocated; 4.56 MiB free; 22.06 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

---

1.AttributeError: module 'numpy' has no attribute 'int'

解决方法：AttributeError: module numpy has no attribute int .报错解决方案_小恶魔饿了的博客-CSDN博客

2.RuntimeError: result type Float can't be cast to the desired output type long int

解决方法：

解决错误：RuntimeError: result type Float can‘t be cast to the desired output type __int64_result type float can't be cast to the desired out_顾悦西的博客-CSDN博客

3.运行很慢很慢，gpu占用率低低低

解决方法：

YOLOv5训练速度慢 GPU占用率低_Goyavae的博客-CSDN博客

会快一点，gpu使用率明显高了一点（呜呜，也没有高很多）

我还是觉得不够快，于是：

使用GPU训练深度模型时，GPU显存占用低，利用率显示为0，而CPU占用很高。请问这怎么解决呢？ - 知乎 (zhihu.com)

我把batch-size调整了，调整到100和我的epochs一样，结果出错：

4.OutOfMemoryError: CUDA out of memory. Tried to allocate 50.00 MiB (GPU 0; 23.65 GiB total capacity; 21.91 GiB already allocated; 4.56 MiB free; 22.06 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

解决办法：

训练yolov5时，RuntimeError: CUDA out of memory._yolo memoryerror_口在天上，数在心中的博客-CSDN博客

于是我又把batch-size改小到50，不行；改到30还是不行；我改到10终于就可以了，也要更快一些，占用gpu也要更多一些

参考：

深度学习训练中的GPU利用率和显存占用问题、num_workers&batch_size设置问题 - 知乎 (zhihu.com)

---