caffe.md

Caffe

Caffe训练模型可视化

caffe模型训练可视化

记录训练日志

训练阶段需要加上 -log_dir ./log/ , 其中 ./log/ 为log文件存放文件文件夹：

~/caffe/build/tools/caffe train --solver=~/caffe/examples/mydata/slot_classifier/solver.prototxt -log_dir ./log/

解析训练日志

将 caffe/tools/extra 文件夹下的 extract_seconds.py , parse_log.py , parse_log.sh, plot_training_log.py.example拷贝到上述的./log/文件夹下.

分步法

1. 修改日志文件名删除caffe.hostname.username.log之后的.INFO.XXXX，保存为caffe.hostname.username.log文件 创建软连接 ln -s caffe.hostname.username.log.INFO.time caffe.log (hostname和username具体根据个人电脑, 下面依然)；
2. 执行: ./parse_log.sh caffe.log , 这样就会在当前文件夹下生成一个.train文件和一个.test文件;
3. 执行:

   ./plot_training_log.py.example 0  save.png caffe.log

   就可以生成训练过程中的Test accuracy vs. Iters 曲线,其中0代表曲线类型， save.png 代表保存的图片名称, caffe中支持很多种曲线绘制，通过指定不同的类型参数即可，具体参数如下:

   Notes:
      1. Supporting multiple logs.
      2. Log file name must end with the lower-cased ".log".
   Supported chart types:
       0: Test accuracy  vs. Iters
       1: Test accuracy  vs. Seconds
       2: Test loss  vs. Iters
       3: Test loss  vs. Seconds
       4: Train learning rate  vs. Iters
       5: Train learning rate  vs. Seconds
       6: Train loss  vs. Iters
       7: Train loss  vs. Seconds

一步法

1. 创建软连接 ln -s caffe.hostname.username.log.INFO.time caffe.log ( hostname 和 username 具体根据个人电脑, 下面依然)；
2. 运行:

    ./plot_training_log.py  [数字选项] 图片名.png ./caffe.log

   其中数字选项如下:

    Notes:
       1. Supporting multiple logs.
       2. Log file name must end with the lower-cased ".log".
    Supported chart types:
        0: Test accuracy  vs. Iters
        1: Test accuracy  vs. Seconds
        2: Test loss  vs. Iters
        3: Test loss  vs. Seconds
        4: Train learning rate  vs. Iters
        5: Train learning rate  vs. Seconds
        6: Train loss  vs. Iters
        7: Train loss  vs. Seconds