# Explicit Access to the Training Set During scVI Training

**URL:** <https://discourse.scverse.org/t/explicit-access-to-the-training-set-during-scvi-training/2137>\
**Category:** scvi-tools\
**Tags:** integration, scvi\
**Created:** [March 4, 2024, 10:50am UTC](https://discourse.scverse.org/t/explicit-access-to-the-training-set-during-scvi-training/2137 "2024-03-04T10:50:08Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![g12bftd](https://avatars.discourse-cdn.com/v4/letter/g/bcef8e/32.png) [@g12bftd](https://discourse.scverse.org/u/g12bftd)\
**Post date:** [March 4, 2024, 10:50am UTC](https://discourse.scverse.org/t/explicit-access-to-the-training-set-during-scvi-training/2137/1 "2024-03-04T10:50:08Z")

</div>

I would like to explicitly access the **training DataLoader** to compute quantities things like Pearson/Spearman gene correlations, gene entropies, NearestNeighbors etc. It is important that these batches are **not computed per batch** , but rather on the **entire training set**. What is the best way to go about this?