首页
开源
资讯
活动
开源许可证
软件工程云服务
软件代码质量检测云服务
持续集成与部署云服务
社区个性化内容推荐服务
贡献审阅人推荐服务
群体化学习服务
重睛鸟代码扫描工具
登录
注册
代码拉取完成,页面将自动刷新
Watch
3
Star
15
Fork
6
MindSpore Lab
/
mindyolo
代码
Issues
27
Pull Requests
1
Wiki
0
统计
更新失败,请稍后重试!
Issues
/
详情
移除标识
内容风险标识
本任务被
标识为内容中包含有代码安全 Bug 、隐私泄露等敏感信息,仓库外成员不可访问
[Bug]: [Block]:训练失败,必现
待办的
#I8VEFS
niannian
创建于
2024-01-11 16:31
### 问题描述 运行命令:python train.py --config ./configs/yolov5/yolov5s.yaml --ckpt_dir /home/ubuntu/cv_lab/yolov5s_300e_mAP376-860bcf3b.ckpt --run_eval True --epochs 300 --per_batch_size 32 --save_dir ./runsEval 训练2个epoch之后自动停止,且精度为0 ### 环境信息 - **Hardware Environment(`Ascend`/`GPU`/`CPU`) / 硬件环境**: 昇腾910B > Please delete the backend not involved / 请删除不涉及的后端:Ascend > /device Ascend - **Software Environment / 软件环境 (Mandatory / 必填)**: -- MindSpore version (e.g., 2.0.0) :2.2.0 -- Python version (e.g., Python 3.7.5) :3.9 -- OS platform and distribution (e.g., Linux Ubuntu 16.04):20.0 -- GCC/Compiler version (if compiled from source): - **Execute Mode / 执行模式 (Mandatory / 必填)(`PyNative`/`Graph`)**: > Please delete the mode not involved / 请删除不涉及的模式:None > /mode pynative > /mode graph ### 关联用例 python train.py --config ./configs/yolov5/yolov5s.yaml --ckpt_dir /home/ubuntu/cv_lab/yolov5s_300e_mAP376-860bcf3b.ckpt --run_eval True --epochs 300 --per_batch_size 32 --save_dir ./runsEval ### 重现步骤 1:准备好训练数据 2:修改配置文件。yolov5s.yaml 3:运行命令:python train.py --config ./configs/yolov5/yolov5s.yaml --ckpt_dir /home/ubuntu/cv_lab/yolov5s_300e_mAP376-860bcf3b.ckpt --run_eval True --epochs 300 --per_batch_size 32 --save_dir ./runsEval 4:报错:2024-01-11 08:17:43,195 [INFO] Speed: 4.4/499.0/503.3 ms inference/NMS/total per 640x640 image at batch-size 32; 2024-01-11 08:17:43,377 [INFO] Epoch 2/300, eval accuracy: 0.000, run_eval time: 76.570 s. 2024-01-11 08:17:43,378 [INFO] best accuracy: 0.000, saved at: ./runsEval/2024.01.11-08.04.44/weights/best_yolov5s-1_17_acc0.000.ckpt Segmentation fault (core dumped) (mindspore_py39) root@ubuntu:/home/ubuntu/cv_lab/code/mindyolo# [WARNING] ME(1195489:281473325449248,WriterPool-29):2024-01-11-08:17:45.284.768 [mindspore/train/summary/_writer_pool.py:193] The training process 1194027 has exited, summary process will exit. /home/envs/miniconda/envs/mindspore_py39/lib/python3.9/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 100 leaked semaphore objects to clean up at shutdown warnings.warn('resource_tracker: There appear to be %d ' ^C (mindspore_py39) root@ubuntu:/home/ubuntu/cv_lab/code/mindyolo# (mindspore_py39) root@ubuntu:/home/ubuntu/cv_lab/code/mindyolo# **粗体** **粗体** _强调_ ### 预期结果 预期训练正常,精度不为0 ### 日志/截图 [在这里上传图片] 验证精度为0 ![](https://foruda.gitee.com/images/1704961842323484446/43f7eb24_5091439.png "屏幕截图") 训练失败 ![输入图片说明](https://foruda.gitee.com/images/1704961873864576948/993d7211_5091439.png "屏幕截图") ### 备注
### 问题描述 运行命令:python train.py --config ./configs/yolov5/yolov5s.yaml --ckpt_dir /home/ubuntu/cv_lab/yolov5s_300e_mAP376-860bcf3b.ckpt --run_eval True --epochs 300 --per_batch_size 32 --save_dir ./runsEval 训练2个epoch之后自动停止,且精度为0 ### 环境信息 - **Hardware Environment(`Ascend`/`GPU`/`CPU`) / 硬件环境**: 昇腾910B > Please delete the backend not involved / 请删除不涉及的后端:Ascend > /device Ascend - **Software Environment / 软件环境 (Mandatory / 必填)**: -- MindSpore version (e.g., 2.0.0) :2.2.0 -- Python version (e.g., Python 3.7.5) :3.9 -- OS platform and distribution (e.g., Linux Ubuntu 16.04):20.0 -- GCC/Compiler version (if compiled from source): - **Execute Mode / 执行模式 (Mandatory / 必填)(`PyNative`/`Graph`)**: > Please delete the mode not involved / 请删除不涉及的模式:None > /mode pynative > /mode graph ### 关联用例 python train.py --config ./configs/yolov5/yolov5s.yaml --ckpt_dir /home/ubuntu/cv_lab/yolov5s_300e_mAP376-860bcf3b.ckpt --run_eval True --epochs 300 --per_batch_size 32 --save_dir ./runsEval ### 重现步骤 1:准备好训练数据 2:修改配置文件。yolov5s.yaml 3:运行命令:python train.py --config ./configs/yolov5/yolov5s.yaml --ckpt_dir /home/ubuntu/cv_lab/yolov5s_300e_mAP376-860bcf3b.ckpt --run_eval True --epochs 300 --per_batch_size 32 --save_dir ./runsEval 4:报错:2024-01-11 08:17:43,195 [INFO] Speed: 4.4/499.0/503.3 ms inference/NMS/total per 640x640 image at batch-size 32; 2024-01-11 08:17:43,377 [INFO] Epoch 2/300, eval accuracy: 0.000, run_eval time: 76.570 s. 2024-01-11 08:17:43,378 [INFO] best accuracy: 0.000, saved at: ./runsEval/2024.01.11-08.04.44/weights/best_yolov5s-1_17_acc0.000.ckpt Segmentation fault (core dumped) (mindspore_py39) root@ubuntu:/home/ubuntu/cv_lab/code/mindyolo# [WARNING] ME(1195489:281473325449248,WriterPool-29):2024-01-11-08:17:45.284.768 [mindspore/train/summary/_writer_pool.py:193] The training process 1194027 has exited, summary process will exit. /home/envs/miniconda/envs/mindspore_py39/lib/python3.9/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 100 leaked semaphore objects to clean up at shutdown warnings.warn('resource_tracker: There appear to be %d ' ^C (mindspore_py39) root@ubuntu:/home/ubuntu/cv_lab/code/mindyolo# (mindspore_py39) root@ubuntu:/home/ubuntu/cv_lab/code/mindyolo# **粗体** **粗体** _强调_ ### 预期结果 预期训练正常,精度不为0 ### 日志/截图 [在这里上传图片] 验证精度为0 ![](https://foruda.gitee.com/images/1704961842323484446/43f7eb24_5091439.png "屏幕截图") 训练失败 ![输入图片说明](https://foruda.gitee.com/images/1704961873864576948/993d7211_5091439.png "屏幕截图") ### 备注
评论 (
0
)
niannian
创建了
任务
1年前
登录
后才可以发表评论
状态
待办的
待办的
进行中
已完成
已关闭
负责人
未设置
标签
未设置
标签管理
里程碑
未关联里程碑
未关联里程碑
Pull Requests
未关联
未关联
关联的 Pull Requests 被合并后可能会关闭此 issue
分支
未关联
分支 (6)
标签 (4)
master
gh-pages
v0.4.0
v0.3.0
v0.2.0
v0.1.0
v0.3
v0.2
v0.1
v0.0.1-alpha
开始日期 - 截止日期
-
置顶选项
不置顶
不置顶
置顶等级:高
置顶等级:中
置顶等级:低
优先级
不指定
不指定
严重
主要
次要
不重要
参与者(1)
Fork 仓库
加载中
取消
确认
UploadFile
FileDragTip
取消
插入