# About Lora Training Script

**URL:** <https://discuss.huggingface.co/t/about-lora-training-script/38437>\
**Category:** 🧨 Diffusers\
**Created:** [May 2, 2023, 2:55pm UTC](https://discuss.huggingface.co/t/about-lora-training-script/38437 "2023-05-02T14:55:12Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![demegire](https://avatars.discourse-cdn.com/v4/letter/d/96bed5/32.png) [@demegire](https://discuss.huggingface.co/u/demegire)\
**Post date:** [May 2, 2023, 2:55pm UTC](https://discuss.huggingface.co/t/about-lora-training-script/38437/1 "2023-05-02T14:55:12Z")

</div>

Hello,

I want to use [diffusers/train\_dreambooth\_lora.py at main · huggingface/diffusers · GitHub](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora.py). However I am not sure what ‘instance\_prompt’ and ‘class\_prompt’ is. What is their difference and how is the logic of a ‘class’ is used in this context?

Also do I have to name the images in my dataset like ‘dog in the style of xyz.png’? I have seen some tutorials do this and some skip this. Would naming them help the training?

---

<div class="post-metadata">

**Author:** ![williamberman](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/williamberman/32/14112_2.png) [@williamberman](https://discuss.huggingface.co/u/williamberman)\
**Post date:** [May 2, 2023, 8:39pm UTC](https://discuss.huggingface.co/t/about-lora-training-script/38437/2 "2023-05-02T20:39:46Z")

</div>

hey @demegire!

the instance prompt and class prompt is documented in the README

> <https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/README.md>

> <https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/README.md#training-with-prior-preservation-loss>

I don’t know off the top of my head if giving additional captions to your dataset will help or not. We don’t have additional prompts for our example dataset. Afaik, there’s not consensus rules on it. I recomend experimenting