我一直在使用单GPU(GeForce GTX 980 Ti,6GB)的普通机器上训练Inception V3.最大批量大小似乎在附近40.
我用在指定的默认学习速率设置inception_train.py文件:initial_learning_rate = 0.1,num_epochs_per_decay = 30和learning_rate_decay_factor = 0.16.经过几周的培训,我能够达到的最佳准确度如下(约500K-1M次迭代):
2016-06-06 12:07:52.245005: precision @ 1 = 0.5767 recall @ 5 = 0.8143 [50016 examples]
2016-06-09 22:35:10.118852: precision @ 1 = 0.5957 recall @ 5 = 0.8294 [50016 examples]
2016-06-14 15:30:59.532629: precision @ 1 = 0.6112 recall @ 5 = 0.8396 [50016 examples]
2016-06-20 13:57:14.025797: precision @ 1 = 0.6136 recall @ 5 = 0.8423 [50016 examples]
我已经尝试在训练课程结束时摆弄设置,但是在准确性方面没有看到任何改进.
我已经从头开始了一个新的训练课程,基于这个论坛中的其他一些帖子, …