# Max number of images in training

**URL:** <https://discuss.pytorch.org/t/max-number-of-images-in-training/123996>\
**Category:** vision\
**Created:** [June 13, 2021, 9:16am UTC](https://discuss.pytorch.org/t/max-number-of-images-in-training/123996 "2021-06-13T09:16:51Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![sj\_xu](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/sj_xu/32/26521_2.png) [@sj\_xu](https://discuss.pytorch.org/u/sj_xu)\
**Post date:** [June 13, 2021, 9:16am UTC](https://discuss.pytorch.org/t/max-number-of-images-in-training/123996/1 "2021-06-13T09:16:51Z")

</div>

Hi, I am having difficulties in using dataloader. I want to train the network on ImageNet with a fixed number of training images. i.e. I hope to set max\_images\_train during the training. But I don’t know how to achieve this in pytorch. Thanks!

---

<div class="post-metadata">

**Author:** ![eqy](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/eqy/32/46233_2.png) [@eqy](https://discuss.pytorch.org/u/eqy)\
**Post date:** [June 13, 2021, 9:47pm UTC](https://discuss.pytorch.org/t/max-number-of-images-in-training/123996/2 "2021-06-13T21:47:27Z")

</div>

Are you trying to limit the number of unique images that are used? You can probably do this by editing the `DatasetFolder` dataloader:  
[https://pytorch.org/vision/stable/\_modules/torchvision/datasets/folder.html#datasetfolder](https://pytorch.org/vision/stable/_modules/torchvision/datasets/folder.html#datasetfolder)  
or by simply creating a modified dataset on disk with some of the images removed.

---

<div class="post-metadata">

**Author:** ![sj\_xu](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/sj_xu/32/26521_2.png) [@sj\_xu](https://discuss.pytorch.org/u/sj_xu)\
**Post date:** [June 13, 2021, 10:04pm UTC](https://discuss.pytorch.org/t/max-number-of-images-in-training/123996/3 "2021-06-13T22:04:14Z")

</div>

Yes, that’s what I want to do. Thank you for your reply, I will try that.