search

huangzehao / caffe-vdsr

A Caffe-based implementation of very deep convolution network for image super-resolution
MIT License
273 stars 134 forks source link

question about training time #8

Closed BaiYu0120 closed 8 years ago

BaiYu0120 commented 8 years ago

hi huang, I am going to training the x4 models using your vdsr training code and 91 images database, how long time will I take? It has been running for about 16 hours.

huangzehao commented 8 years ago

Hi, it depends on your gpus. You can firstly check the performance of your snapshots and decide when to stop. In my experiment, model trained about 30 epoch can get a good result.

zhangfudiyi commented 8 years ago

@huangzehao 你好，请问一下1个epoch是迭代多少次？

huangzehao commented 8 years ago

@zhangfudiyi It depends on the number of training samples and batch size. Iters_num of one epoch = training_num / batch_size

zhangfudiyi commented 8 years ago

@huangzehao thank you！