Dragon下跑Cifar10 example

最新推荐文章于 2022-07-25 22:55:29 发布

qq_27278153

最新推荐文章于 2022-07-25 22:55:29 发布

阅读量4.9k

点赞数

分类专栏： Dragon

本文链接：https://blog.csdn.net/qq_27278153/article/details/78450498

版权

Dragon 专栏收录该内容

2 篇文章 0 订阅

订阅专栏

虽然配置过了Dragon，但是根据Caffe的经验，还是要能够先跑通一个example，才能确定确实是安装成功了。才更放心。

　　主要参考了：

　　我的Pycharm，我做主

　　Dragon/examples/cifar10/

　　整体总结：

1）官方钦点了Pycharm，所以就跟着撸起来。

2）

3）

4）

5）

　　在windows下，就是直接复制，然后在pycharm里面，直接运行 gen_lmdb.py

　　然后再在pycharm里面运行了solve_quick.py

C:\Anaconda2\python.exe C:/Dragon/Dragon-master/examples/cifar10/solve_quick.py
---------------------------------------------------------
BatchReader, Using config:
{'num_fetchers': 1, 'num_readers': 1, 'num_transformers': 1, 'prefetching': 5}
---------------------------------------------------------
---------------------------------------------------------
BatchReader, Using config:
{'num_fetchers': 1, 'num_readers': 1, 'num_transformers': 1, 'prefetching': 5}
---------------------------------------------------------
---------------------------------------------------------
Optimizer: SGD, Using config:
{'base_lr': 0.001,
 'clip_gradient': -1.0,
 'l2_decay': 0.004,
 'momentum': 0.9,
 'scale_gradient': 1.0}
---------------------------------------------------------
Iteration 0, lr = 0.001, loss = 2.302690, time = 0.04s
      Train net output #0(loss): 2.30269002914
Iteration 100, lr = 0.001, loss = 1.646297, time = 1.21s
      Train net output #0(loss): 1.64629745483
Iteration 200, lr = 0.001, loss = 1.689907, time = 1.28s
      Train net output #0(loss): 1.68990671635
Iteration 300, lr = 0.001, loss = 1.363599, time = 1.21s
      Train net output #0(loss): 1.36359858513
Iteration 400, lr = 0.001, loss = 1.295662, time = 1.19s
      Train net output #0(loss): 1.29566156864
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_500.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 500, Test net #0
       Test net output #0(loss): 1.3241
       Test net output #1(accuracy): 0.5337
Iteration 500, lr = 0.001, loss = 1.306685, time = 2.30s
      Train net output #0(loss): 1.30668485165
Iteration 600, lr = 0.001, loss = 1.248834, time = 1.07s
      Train net output #0(loss): 1.24883377552
Iteration 700, lr = 0.001, loss = 1.331041, time = 1.19s
      Train net output #0(loss): 1.3310406208
Iteration 800, lr = 0.001, loss = 1.082184, time = 1.19s
      Train net output #0(loss): 1.08218431473
Iteration 900, lr = 0.001, loss = 1.055118, time = 1.19s
      Train net output #0(loss): 1.05511772633
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_1000.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 1000, Test net #0
       Test net output #0(loss): 1.0990
       Test net output #1(accuracy): 0.6170
Iteration 1000, lr = 0.001, loss = 1.100702, time = 2.26s
      Train net output #0(loss): 1.10070216656
Iteration 1100, lr = 0.001, loss = 0.989666, time = 1.06s
      Train net output #0(loss): 0.989666223526
Iteration 1200, lr = 0.001, loss = 1.148743, time = 1.19s
      Train net output #0(loss): 1.14874303341
Iteration 1300, lr = 0.001, loss = 1.007544, time = 1.19s
      Train net output #0(loss): 1.00754368305
Iteration 1400, lr = 0.001, loss = 0.979396, time = 1.19s
      Train net output #0(loss): 0.979396104813
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_1500.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 1500, Test net #0
       Test net output #0(loss): 1.0430
       Test net output #1(accuracy): 0.6443
Iteration 1500, lr = 0.001, loss = 0.993701, time = 2.33s
      Train net output #0(loss): 0.993700861931
Iteration 1600, lr = 0.001, loss = 0.940067, time = 1.16s
      Train net output #0(loss): 0.940066814423
Iteration 1700, lr = 0.001, loss = 1.057620, time = 1.29s
      Train net output #0(loss): 1.05762040615
Iteration 1800, lr = 0.001, loss = 0.901356, time = 1.25s
      Train net output #0(loss): 0.901355564594
Iteration 1900, lr = 0.001, loss = 0.889179, time = 1.23s
      Train net output #0(loss): 0.889179170132
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_2000.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 2000, Test net #0
       Test net output #0(loss): 0.9478
       Test net output #1(accuracy): 0.6823
Iteration 2000, lr = 0.001, loss = 0.872262, time = 2.39s
      Train net output #0(loss): 0.872261941433
Iteration 2100, lr = 0.001, loss = 0.853221, time = 1.11s
      Train net output #0(loss): 0.853220760822
Iteration 2200, lr = 0.001, loss = 0.979079, time = 1.22s
      Train net output #0(loss): 0.979078531265
Iteration 2300, lr = 0.001, loss = 0.880974, time = 1.19s
      Train net output #0(loss): 0.880973517895
Iteration 2400, lr = 0.001, loss = 0.847923, time = 1.19s
      Train net output #0(loss): 0.847923219204
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_2500.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 2500, Test net #0
       Test net output #0(loss): 0.9196
       Test net output #1(accuracy): 0.6936
Iteration 2500, lr = 0.001, loss = 0.808091, time = 2.32s
      Train net output #0(loss): 0.80809122324
Iteration 2600, lr = 0.001, loss = 0.862147, time = 1.10s
      Train net output #0(loss): 0.862147092819
Iteration 2700, lr = 0.001, loss = 0.866439, time = 1.20s
      Train net output #0(loss): 0.866439223289
Iteration 2800, lr = 0.001, loss = 0.819624, time = 1.19s
      Train net output #0(loss): 0.819624185562
Iteration 2900, lr = 0.001, loss = 0.805170, time = 1.20s
      Train net output #0(loss): 0.805170238018
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_3000.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 3000, Test net #0
       Test net output #0(loss): 0.8814
       Test net output #1(accuracy): 0.7082
Iteration 3000, lr = 0.001, loss = 0.721643, time = 2.23s
      Train net output #0(loss): 0.721643269062
Iteration 3100, lr = 0.001, loss = 0.846673, time = 1.10s
      Train net output #0(loss): 0.846673071384
Iteration 3200, lr = 0.001, loss = 0.820102, time = 1.19s
      Train net output #0(loss): 0.82010191679
Iteration 3300, lr = 0.001, loss = 0.768485, time = 1.18s
      Train net output #0(loss): 0.768485248089
Iteration 3400, lr = 0.001, loss = 0.745497, time = 1.20s
      Train net output #0(loss): 0.74549728632
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_3500.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 3500, Test net #0
       Test net output #0(loss): 0.8588
       Test net output #1(accuracy): 0.7153
Iteration 3500, lr = 0.001, loss = 0.646889, time = 2.24s
      Train net output #0(loss): 0.646888554096
Iteration 3600, lr = 0.001, loss = 0.805500, time = 1.12s
      Train net output #0(loss): 0.805499851704
Iteration 3700, lr = 0.001, loss = 0.768448, time = 1.21s
      Train net output #0(loss): 0.76844805479
Iteration 3800, lr = 0.001, loss = 0.650282, time = 1.21s
      Train net output #0(loss): 0.650282263756
Iteration 3900, lr = 0.001, loss = 0.695232, time = 1.20s
      Train net output #0(loss): 0.69523203373
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_4000.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 4000, Test net #0
       Test net output #0(loss): 0.8711
       Test net output #1(accuracy): 0.7127
Iteration 4000, lr = 0.0001, loss = 0.580074, time = 2.26s
      Train net output #0(loss): 0.580074310303
Iteration 4100, lr = 0.0001, loss = 0.652031, time = 1.13s
      Train net output #0(loss): 0.652031123638
Iteration 4200, lr = 0.0001, loss = 0.626367, time = 1.23s
      Train net output #0(loss): 0.626367032528
Iteration 4300, lr = 0.0001, loss = 0.544869, time = 1.21s
      Train net output #0(loss): 0.544868648052
Iteration 4400, lr = 0.0001, loss = 0.499933, time = 1.20s
      Train net output #0(loss): 0.499933123589
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_4500.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Iteration 4500, Test net #0
       Test net output #0(loss): 0.7643
       Test net output #1(accuracy): 0.7460
Iteration 4500, lr = 0.0001, loss = 0.562213, time = 2.25s
      Train net output #0(loss): 0.562212586403
Iteration 4600, lr = 0.0001, loss = 0.631839, time = 1.14s
      Train net output #0(loss): 0.63183927536
Iteration 4700, lr = 0.0001, loss = 0.598695, time = 1.20s
      Train net output #0(loss): 0.598694562912
Iteration 4800, lr = 0.0001, loss = 0.528736, time = 1.19s
      Train net output #0(loss): 0.528735756874
Iteration 4900, lr = 0.0001, loss = 0.481992, time = 1.19s
      Train net output #0(loss): 0.481991797686
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_5000.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
INFO utils/caffemodel.h:114] Save the model @: snapshots/cifar10_quick_iter_5000.caffemodel......
INFO utils/caffemodel.h:115] Model format: caffemodel
Terminating BlobFetcher......
Terminating DataTransformer......
Terminating DataReader......
Terminating BlobFetcher......
Terminating DataTransformer......
Terminating DataReader......

Process finished with exit code 0

　　运行成功。说明Dragon安装成功。

　　跟着也跑了一下。

C:\Anaconda2\python.exe C:/Dragon/Dragon-master/examples/cifar10/infer.py
INFO utils/caffemodel.h:55] Restore From Model @: snapshots/cifar10_quick_iter_5000.caffemodel......
INFO utils/caffemodel.h:56] Model Format: CaffeModel
INFO utils/caffemodel.h:86] Tensor(conv1@param0) loaded, shape: (32,3,5,5), size: 2400
INFO utils/caffemodel.h:86] Tensor(conv1@param1) loaded, shape: (32), size: 32
INFO utils/caffemodel.h:86] Tensor(conv2@param0) loaded, shape: (32,32,5,5), size: 25600
INFO utils/caffemodel.h:86] Tensor(conv2@param1) loaded, shape: (32), size: 32
INFO utils/caffemodel.h:86] Tensor(conv3@param0) loaded, shape: (64,32,5,5), size: 51200
INFO utils/caffemodel.h:86] Tensor(conv3@param1) loaded, shape: (64), size: 64
INFO utils/caffemodel.h:86] Tensor(ip1@param0) loaded, shape: (64,1024), size: 65536
INFO utils/caffemodel.h:86] Tensor(ip1@param1) loaded, shape: (64), size: 64
INFO utils/caffemodel.h:86] Tensor(ip2@param0) loaded, shape: (10,64), size: 640
INFO utils/caffemodel.h:86] Tensor(ip2@param1) loaded, shape: (10), size: 10
cat

Process finished with exit code 0

qq_27278153

关注

0
点赞
踩
0

收藏

觉得还不错? 一键收藏
0
评论
Dragon下跑Cifar10 example

虽然配置过了Dragon，但是根据Caffe的经验，还是要能够先跑通一个example，才能确定确实是安装成功了。才更放心。　　主要参考了：　　我的Pycharm，我做主　　Dragon/examples/cifar10/　　整体总结：1）官方钦点了Pycharm，所以就跟着撸起来。2）3）4）5）　　　　
复制链接

扫一扫