Replies: 5 comments
-
补充一下,我是训练ser模型任务,执行命令是:! python3 -m paddle.distributed.launch --gpus '0' tools/train.py -c configs/kie/vi_layoutxlm/ser_vi_layoutxlm_xfund_zh.yml,也下载了官方文档提供的预训练模型 |
Beta Was this translation helpful? Give feedback.
-
你好,之前确实没有遇到过这类情况,可以问一下这个问题是可以复现的吗,还是说是偶然发生? |
Beta Was this translation helpful? Give feedback.
-
感觉是官方的预训练模型有问题,我是在win10下跑的,下载的预训练模型tar文件,解压后得到的其实还是个压缩包,修改成.tar或者是.rar后缀再解压一次会得到3个文件,里面有模型model_state.pdparams,但是缺了个.pdopt,所以一般方法直接使用breakpoints导入是用不了的 |
Beta Was this translation helpful? Give feedback.
-
根据您提供的信息以及 PaddleOCR 的配置和问题描述,以下是关于训练中 1. 数据集问题
2. 训练配置问题
3. 模型架构与输入尺寸的适配性
4. 后处理与评估问题
5. 调试与诊断建议
6. 尝试官方建议
建议的调整示例以下是根据您的情况修改后的部分配置: Train:
dataset:
name: SimpleDataSet
data_dir: C:\Users\Czw\Desktop\ocr\det/
label_file_list:
- C:\Users\Czw\Desktop\ocr\det\train.txt
transforms:
- DecodeImage:
img_mode: BGR
channel_first: false
- DetLabelEncode: null
- Resize:
size:
- 400 # 调整为更适合数据的尺寸
- 64
- NormalizeImage:
scale: 1./255.
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
order: hwc
- ToCHWImage: null
- KeepKeys:
keep_keys:
- image
- threshold_map
- threshold_mask
- shrink_map
- shrink_mask
loader:
shuffle: true
drop_last: false
batch_size_per_card: 4
num_workers: 2 总结
如果问题仍未解决,建议提供更多详细日志或最小可复现示例,便于进一步诊断。 Response generated by feifei-bot | chatgpt-4o-latest |
Beta Was this translation helpful? Give feedback.
-
我也遇到到,标签问题,标签是关键字,相当于是都识别成other了 |
Beta Was this translation helpful? Give feedback.
-
Search before asking
I have searched the PaddleOCR Docs and found no similar bug report.
I have searched the PaddleOCR Issues and found no similar bug report.
I have searched the PaddleOCR Discussions and found no similar bug report.
Bug
你好,我司正在调研百度飞桨的ppstructure的kie模块,我是在百度的aistudio平台训练kie的模型,参考官方手册一步一步进行的。问题是如标题所示,79轮前面的第78轮都还是正常的,到79轮及以后,cur metric, precision: 0.0, recall: 0.0, hmean: 0.0 三个参数都为0 ,不清楚什么原因。我贴一下截图
请麻烦解答一下,谢谢!我可以提供我的数据集
Environment
No response
Minimal Reproducible Example
No response
Additional
No response
Are you willing to submit a PR?
Beta Was this translation helpful? Give feedback.
All reactions