Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

请教下,自己训练遇到的疑问 #111

Open
zkailinzhang opened this issue Feb 2, 2018 · 3 comments
Open

请教下,自己训练遇到的疑问 #111

zkailinzhang opened this issue Feb 2, 2018 · 3 comments

Comments

@zkailinzhang
Copy link

  1. 遇到的第一个问题是 data文件夹位置没放好,要放在Wechat_AutoJump目录下,'E:\python\weixin\Wechat_AutoJump\data',,我之前是放在resource目录下,出现问题,即时修改两个文件 jump_data_fine.py jump_data.py

  2. 接着我仅仅是在base.fine中跑 train.py, 8个小时吧,出现train_logs文件夹 包含 7个文件,两个模型吧,以及现在的训练状态和误差
    _20180202104934
    _20180202105304

训练中会看到这样的信息 每一千次吧 val_loss, best_val_loss,,而大部分batch train_loss,,大神能讲下这两组参数的区别的吗

_20180202095537

3.大神在之前评论说过,cnn的优势是zai在 在16:9的数据上训好的cnn在非16:9下也能work,,所以也自己尝试去训练,但是训练好的模型,要怎么处理,不是太明白,
目标是要在resource目录下放两个文件夹,train_logs_coarse train_logs_fine,

  • 现在我只是训练了 config/base.fine,,我还要再训练config/base.large,是吗,然后再选这两个文件夹的最新的模型cp过去

  • 不训练config/base.large,可以吗,用楼主提供的,

  • 再者config/base.fine也生成了两个模型,999 1999,,一个放fine 一个放coarse,,不是这样理解的吧

  1. 最后一个疑问,就是 play_nn.py为啥要喂两个模型,一个不行吗,直接fine,,提前谢谢大神
@leon332157
Copy link
Contributor

@Prinsphield @RICHARD-AN

@zkailinzhang
Copy link
Author

另外我再问下, config\base.fine\train_logs下的模型,多少次bacth会生成新的模型,昨天晚上,生成best_model.ckpt-999和best_model.ckpt-1999,到现在还没有生成新的model,train_loss现在是4.0,,我怎么拿到最新的模型呢,

@Prinsphield
Copy link
Owner

base.fine是fine model,base.large是coarse model,coarse model应该可以用传统方式替换掉,根据小人的位置在图中找到对称的位置,然后框一个320*320的框,然后直接训练fine model。直接用一个网络来学原图大小下的target position,可能不是很准,因为图片太大。

end to end的做法是输入是原图,输出是press time,这个有别的同学已经做了。你可以用我训练好的模型跑,自己造一些数据集出来用作end to end的训练集。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants