---
title: libwaifu-noobai-xl-v1.1
canonical_url: "https://www.modelscope.cn/models/ling0322/libwaifu-noobai-xl-v1.1"
md_url: "https://www.modelscope.cn/models/ling0322/libwaifu-noobai-xl-v1.1.md"
repository: ling0322/libwaifu-noobai-xl-v1.1
last_updated: 2026-09-22
license: other
pipeline_tag: text-to-image-synthesis
tasks:
  - text-to-image-synthesis
base_model:
  - Laxhar/noobai-XL-1.1
base_model_relation: finetune
parameters: 3.5B
tensor_type:
  - F16
  - F32
library_name:
  - lora
  - safetensors
downloads: 3
stars: 0
tags:
  - text-to-image
  - stable-diffusion
  - stable-diffusion-xl
  - illustrious
  - libwaifu
  - not-for-all-audiences
---

# libwaifu-noobai-xl-v1.1

> libwaifu-noobai-xl-v1.1 - ling0322 在 ModelScope 开源的模型。NoobAI-XL v1.1 for libwaifu

ling0322/libwaifu-noobai-xl-v1.1 是 ModelScope 魔搭社区上的 3.5B 参数text-to-image-synthesis模型，采用 other 许可，基于 Laxhar/noobai-XL-1.1 构建。

- **Repository**: ling0322/libwaifu-noobai-xl-v1.1
- **License**: other
- **Tasks**: text-to-image-synthesis
- **Parameters**: 3.5B
- **Base model**: Laxhar/noobai-XL-1.1
- **Tags**: text-to-image, stable-diffusion, stable-diffusion-xl, illustrious, libwaifu, not-for-all-audiences
- **Downloads**: 3
- **Stars**: 0
- **Last updated**: 2026-09-22

Source: https://www.modelscope.cn/models/ling0322/libwaifu-noobai-xl-v1.1

---

# NoobAI-XL v1.1 for libwaifu

NoobAI-XL v1.1 converted to the format [libwaifu](https://github.com/ling0322/libwaifu) reads, so
it can draw from it directly. The weights are the original ones: the U-Net and both text encoders
are stored as float16, and the VAE is left wider because it overflows float16.

Converted from [Laxhar/noobai-XL-1.1](https://huggingface.co/Laxhar/noobai-XL-1.1), sha256
`6681e8e4b134c81f16533acedb0d406d7e5e366e1624b4105178c64d00b05d51`. It is trained from
Illustrious XL on Danbooru and e621, so it is prompted with danbooru tags rather than with
sentences.

This is the epsilon-prediction release. NoobAI also publishes a v-prediction model, which is a
different parameterization of the same schedule and is **not** what this repository holds --
libwaifu's Euler sampler reads epsilon.

Both halves of the autoencoder are here, so it draws from a picture as well as from a prompt.

## Files

A model is a `.yaml` and the files it names, all of them in one directory. The manifest is a small
text file you can read and edit; the weights are ordinary safetensors, readable by torch, numpy or
the viewer on this page; the tokenizer is the `tokenizer.json` Stability published.

| file | size | what |
|---|---|---|
| `noobai-xl-v1.1.yaml` | 2 KB | the manifest: what the model is, and which files it is made of |
| `noobai-xl-v1.1.tokenizer.json` | 2.1 MiB | the CLIP tokenizer |
| `noobai-xl-v1.1-00001-of-00004.safetensors` | 1.88 GiB | |
| `noobai-xl-v1.1-00002-of-00004.safetensors` | 1.88 GiB | |
| `noobai-xl-v1.1-00003-of-00004.safetensors` | 1.87 GiB | |
| `noobai-xl-v1.1-00004-of-00004.safetensors` | 0.99 GiB | |

The split is between tensors, never inside one, so each file is a safetensors in its own right and
can be read and checked alone. The picture the four draw is identical, to the byte, to the one an
unsplit model draws.

**This repository was called `libwaifu-noobai-xl-v11` until the version in its name grew a dot.**
The old address still resolves, and `sdxl:noob:v11` still names this model in `waifu`.

## Usage

`waifu` knows this model as `sdxl:noob`, so there is nothing to download by hand:

```bash
waifu draw -m sdxl:noob
```

`sdxl:noob:v1.1` names this release and keeps meaning it; `sdxl:noob` follows whatever is current.

```bash
hf download ling0322/libwaifu-noobai-xl-v1.1 --local-dir noobai-xl-v1.1
waifu draw -m noobai-xl-v1.1/noobai-xl-v1.1.yaml
```

Keep the files side by side -- a manifest names its neighbours by file name and will not follow a
path anywhere else.

## Prompting

The manifest suggests the author's own prefix, the resolution they mark as best, and the bottom of
the step and guidance ranges they give. What it cannot carry is their negative prompt, so it is
here:

```
nsfw, worst quality, old, early, low quality, lowres, signature, username, logo, bad hands,
mutated hands, mammal, anthro, furry, ambiguous form, feral, semi-anthro
```

Prompts are written as `<1girl/1boy/1other/...>, <character>, <series>, <artists>,
<special tags>, <general tags>, <other tags>`. The author recommends Euler a as the sampler.

## Conversion

```bash
python tools/sdxl_exporter.py \
  -checkpoint NoobAI-XL-v1.1.safetensors \
  -output noobai-xl-v1.1.safetensors \
  -part-size 2GB
```

## License

The weights are the author's; this repository only changes the container they are in. The license
is inherited from the source and is **not** MIT like libwaifu itself: it is
[fair-ai-public-license-1.0-sd](https://freedevproject.org/faipl-1.0-sd/), which prohibits
commercial use and requires that derivative models and the work behind them stay open. See the
[source repository](https://huggingface.co/Laxhar/noobai-XL-1.1) for the full terms and model card.