---
license: apache-2.0
base_model:
- 8BitStudio/Aniimage-2
tags:
- lfm
- portrait
---

UNet

Flow matching (time_shift_type is 'linear') and the [LFM2.5 text encoder](https://huggingface.co/nebulette/clip-l-sized-lfm-230m) on the top of Aniimage.

Due to the lack of training data, the only prompt it understands is anime portraits.

What is more interesting is that the model did not forget all the CLIP embeddings with
<br> the new v-prediction, despite it has been trained only on LFM embeddings. See the image below:

![](images/preview.png)

Source data:
- anime_style_portrait
- gelbooru (landscape)
- portraits_512
- wikiart_face