# One of the biggest challenges in LoRA training is the issue of LoRA weight. Adjusting it—increasing it to 1.1 or decreasing it to 0.9—introduces another variable. Furthermore, there are now new nodes that allow for increasing the weight of specific words — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/e8871a8588f16aab

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-11-14 00:27 ingested 2026-11-15 02:35

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1xvoh0a/one_of_the_biggest_challenges_in_lora_training_is/)

## retained text



```
Should I train my LoRA for longer, or just increase the weight?

Could an overtrained LoRA actually perform better if used at a lower weight?

And what about the weight of specific prompts—especially the trigger word?
```



document fetched 2026-11-15 00:29 · canonical https://reddit.com/r/StableDiffusion/comments/1xvoh0a/one_of_the_biggest_challenges_in_lora_training_is
