Anima — Karume

What is this

The official CircleStone Anima models — circlestone-labs/Anima-Base-v1.0-Diffusers and its official variants — converted into the WebGPU inference runtime Karume's container format (safetensors = weights + a graph JSON embedded in __metadata__, split across numbered shards when a component is too large for one file). Runs as-is in the browser and in Deno.

The default model is anima-turbo-v1.1 — 8 steps / guidance 1 by default. At guidance 1 the second CFG branch is skipped and the negative prompt is not used — pick a model that runs classifier-free guidance (see the model list below) when you want the negative prompt to take effect.
Not readable by diffusers (it's a different container with an embedded graph); the reader is a pipeline that implements anima/1.
Exporter used for the conversion: karume/0.8.0. The distribution manifest is karume.json (karume/4).

Models and their origins

Each model below is an official CircleStone release — the Anima base model itself or an official variant of it. The text encoder, VAE and tokenizers are shared across them.

`anima-turbo-v1.1` — Anima Turbo v1.1

Author: circlestone_labs
Source: https://huggingface.co/circlestone-labs/Anima
Converted from: anima-turbo-v1.1.safetensors

`anima-v1.0` — Anima Base v1.0

Author: circlestone_labs
Source: https://huggingface.co/circlestone-labs/Anima-Base-v1.0-Diffusers

`anima-aesthetic-v1.1` — Anima Aesthetic v1.1

Author: circlestone_labs
Source: https://huggingface.co/circlestone-labs/Anima
Converted from: anima-aesthetic-v1.1.safetensors

`anima-turbo-v1.0` — Anima Turbo v1.0

Author: circlestone_labs
Source: https://huggingface.co/circlestone-labs/Anima
Converted from: anima-turbo-v1.0.safetensors

`anima-aesthetic-v1.0` — Anima Aesthetic v1.0

Author: circlestone_labs
Source: https://huggingface.co/circlestone-labs/Anima
Converted from: anima-aesthetic-v1.0.safetensors

License

Every model here derives from the CircleStone Anima base model and stays under the CircleStone Non-Commercial License (non-commercial use only). This repository ships LICENSE.md (the full license text) and NOTICE.md (this attribution plus the list of modifications).

The CircleStone Model is licensed by CircleStone Labs LLC under the CircleStone Non-Commercial License. Copyright CircleStone Labs LLC. IN NO EVENT SHALL CIRCLESTONE LABS LLC BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

Outputs you generate are yours to use for any purpose, including commercially (license §2(e)); the non-commercial restriction applies to the model weights and derivatives, not to outputs.
This is not an official product of CircleStone Labs LLC, and it is not endorsed, approved or validated by CircleStone Labs LLC.

Models

Model	Pipeline	Quants	Default quant
`anima-turbo-v1.1` (default)	`anima/1`	`f16` / `f16+dit8` / `f16+dit8-a8` / `f16+dit8-a8-attn8` / `f16+dit8-a8-attn8-s16` / `f16-c16`	`f16+dit8-a8-attn8-s16`
`anima-v1.0`	`anima/1`	`f16` / `f16+dit8` / `f16+dit8-a8` / `f16+dit8-a8-attn8` / `f16+dit8-a8-attn8-s16` / `f16-c16`	`f16+dit8-a8-attn8-s16`
`anima-aesthetic-v1.1`	`anima/1`	`f16` / `f16+dit8` / `f16+dit8-a8` / `f16+dit8-a8-attn8` / `f16+dit8-a8-attn8-s16` / `f16-c16`	`f16+dit8-a8-attn8-s16`
`anima-turbo-v1.0`	`anima/1`	`f16` / `f16+dit8` / `f16+dit8-a8` / `f16+dit8-a8-attn8` / `f16+dit8-a8-attn8-s16` / `f16-c16`	`f16+dit8-a8-attn8-s16`
`anima-aesthetic-v1.0`	`anima/1`	`f16` / `f16+dit8` / `f16+dit8-a8` / `f16+dit8-a8-attn8` / `f16+dit8-a8-attn8-s16` / `f16-c16`	`f16+dit8-a8-attn8-s16`

model selects one of these; omitted, it is anima-turbo-v1.1. quant defaults to that model's own default quant.

Usage

import { AnimaPipeline, encodePng } from "jsr:@karume/models";

using pipeline = await AnimaPipeline.fromPretrained({
  repo: "hdae/karume-anima",
  // Pin a commit for reproducible builds — without it you track `main`, and a future
  // repo update (renamed files, new manifest format) may break your app.
  // Copy the full hash from this repo's "Files and versions" tab:
  // revision: "<full commit sha>",
}, {
  // model: "anima-turbo-v1.1", // default — available: anima-aesthetic-v1.0 / anima-aesthetic-v1.1 / anima-turbo-v1.0 / anima-turbo-v1.1 / anima-v1.0
  // quant: "f16+dit8-a8-attn8-s16", // default — available: f16 / f16+dit8 / f16+dit8-a8 / f16+dit8-a8-attn8 / f16+dit8-a8-attn8-s16 / f16-c16
});

const image = await pipeline.generate({
  prompt: "1girl, solo, long hair, blue eyes, school uniform, masterpiece",

  // steps: 8, // default — the model is distilled for few-step sampling
  // Resolution — non-square is fine; each side on a 16 px grid, between 512 and 2048 px:
  // resolution: { width: 1024, height: 1024 }, // default

  // Classifier-free guidance runs a second (uncond) branch — twice the work per step.
  // It is skipped at guidanceScale 1, where a negativePrompt is refused rather than
  // silently ignored, so the two lines below only make sense together:
  // guidanceScale: 5,
  // negativePrompt: "low quality, worst quality, blurry, bad anatomy, jpeg artifacts",

  seed: 42, // same seed + same request → same image
});

const png = await encodePng(image.data, image.width, image.height);
await Deno.writeFile("anima.png", png);

Weights are fetched once and cached (verified against karume.json's size / sha256).

Model: anima-turbo-v1.1

Quants

Quant	What it is	Download	Weights	Compute
`f16`	Full quality (f16) — Transformer in f16 storage with f32 compute — the largest download, and the reference the other quants here are judged against.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	—
`f16+dit8`	Half size (int8 transformer) — Transformer stored as int8 and computed in f32: roughly half its f16 download, with the execution path left unchanged.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	—
`f16+dit8-a8`	Half size, int8 linear — The int8 transformer with per-token int8 activations in its linear layers — faster on GPUs with dp4a, same download.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8`
`f16+dit8-a8-attn8`	Half size, int8 linear and attention — Adds int8 activations inside attention on top of the int8 linear path; same weights, one more integer stage per step.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8`
`f16+dit8-a8-attn8-s16` (default)	Balanced (int8) — The int8 linear and attention path with attention scores held in f16 — the fastest of the int8 seats here, at f16-level image quality.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8` / `attentionScoreStorage` = `f16`
`f16-c16`	Full quality, f16 compute — f16 storage computed in f16 throughout. Needs the shader-f16 GPU feature, and trades numerical headroom for speed.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	`linearCompute` = `f16` / `attentionCompute` = `f16` / requires `shaderF16`

If no quant is given, it runs as f16+dit8-a8-attn8-s16 (this model's recommended default). In a quant name, dit is the transformer component. Per-file size and sha256 live in karume.json — verify against that at the fetch layer. Dtype labels use the runtime's storage dtype vocabulary (f16 / i8 / i4), not the fp16 spelling common elsewhere in the ecosystem. A path under shared/ is one this model shares byte for byte with another model in this repository (it is fetched and cached once).

Defaults

Any knob not passed to generate() is filled in from the manifest's defaults.

steps: 8
guidanceScale: 1
resolution: 1024 × 1024
negativePrompt: low quality, worst quality, blurry, bad anatomy, jpeg artifacts

At guidance 1, the second CFG branch is skipped, so the negative prompt is not used (it only takes effect once guidance is raised).

Model: anima-v1.0

Quants

Quant	What it is	Download	Weights	Compute
`f16`	Full quality (f16) — Transformer in f16 storage with f32 compute — the largest download, and the reference the other quants here are judged against.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	—
`f16+dit8`	Half size (int8 transformer) — Transformer stored as int8 and computed in f32: roughly half its f16 download, with the execution path left unchanged.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	—
`f16+dit8-a8`	Half size, int8 linear — The int8 transformer with per-token int8 activations in its linear layers — faster on GPUs with dp4a, same download.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8`
`f16+dit8-a8-attn8`	Half size, int8 linear and attention — Adds int8 activations inside attention on top of the int8 linear path; same weights, one more integer stage per step.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8`
`f16+dit8-a8-attn8-s16` (default)	Balanced (int8) — The int8 linear and attention path with attention scores held in f16 — the fastest of the int8 seats here, at f16-level image quality.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8` / `attentionScoreStorage` = `f16`
`f16-c16`	Full quality, f16 compute — f16 storage computed in f16 throughout. Needs the shader-f16 GPU feature, and trades numerical headroom for speed.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	`linearCompute` = `f16` / `attentionCompute` = `f16` / requires `shaderF16`

Defaults

Any knob not passed to generate() is filled in from the manifest's defaults.

steps: 20
guidanceScale: 4
resolution: 1024 × 1024
negativePrompt: low quality, worst quality, blurry, bad anatomy, jpeg artifacts

Model: anima-aesthetic-v1.1

Quants

Quant	What it is	Download	Weights	Compute
`f16`	Full quality (f16) — Transformer in f16 storage with f32 compute — the largest download, and the reference the other quants here are judged against.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	—
`f16+dit8`	Half size (int8 transformer) — Transformer stored as int8 and computed in f32: roughly half its f16 download, with the execution path left unchanged.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	—
`f16+dit8-a8`	Half size, int8 linear — The int8 transformer with per-token int8 activations in its linear layers — faster on GPUs with dp4a, same download.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8`
`f16+dit8-a8-attn8`	Half size, int8 linear and attention — Adds int8 activations inside attention on top of the int8 linear path; same weights, one more integer stage per step.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8`
`f16+dit8-a8-attn8-s16` (default)	Balanced (int8) — The int8 linear and attention path with attention scores held in f16 — the fastest of the int8 seats here, at f16-level image quality.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8` / `attentionScoreStorage` = `f16`
`f16-c16`	Full quality, f16 compute — f16 storage computed in f16 throughout. Needs the shader-f16 GPU feature, and trades numerical headroom for speed.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	`linearCompute` = `f16` / `attentionCompute` = `f16` / requires `shaderF16`

Defaults

Any knob not passed to generate() is filled in from the manifest's defaults.

steps: 30
guidanceScale: 4
resolution: 1024 × 1024
negativePrompt: low quality, worst quality, blurry, bad anatomy, jpeg artifacts

Model: anima-turbo-v1.0

Quants

Quant	What it is	Download	Weights	Compute
`f16`	Full quality (f16) — Transformer in f16 storage with f32 compute — the largest download, and the reference the other quants here are judged against.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	—
`f16+dit8`	Half size (int8 transformer) — Transformer stored as int8 and computed in f32: roughly half its f16 download, with the execution path left unchanged.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	—
`f16+dit8-a8`	Half size, int8 linear — The int8 transformer with per-token int8 activations in its linear layers — faster on GPUs with dp4a, same download.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8`
`f16+dit8-a8-attn8`	Half size, int8 linear and attention — Adds int8 activations inside attention on top of the int8 linear path; same weights, one more integer stage per step.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8`
`f16+dit8-a8-attn8-s16` (default)	Balanced (int8) — The int8 linear and attention path with attention scores held in f16 — the fastest of the int8 seats here, at f16-level image quality.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8` / `attentionScoreStorage` = `f16`
`f16-c16`	Full quality, f16 compute — f16 storage computed in f16 throughout. Needs the shader-f16 GPU feature, and trades numerical headroom for speed.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	`linearCompute` = `f16` / `attentionCompute` = `f16` / requires `shaderF16`

Defaults

Any knob not passed to generate() is filled in from the manifest's defaults.

steps: 8
guidanceScale: 1
resolution: 1024 × 1024
negativePrompt: low quality, worst quality, blurry, bad anatomy, jpeg artifacts

At guidance 1, the second CFG branch is skipped, so the negative prompt is not used (it only takes effect once guidance is raised).

Model: anima-aesthetic-v1.0

Quants

Quant	What it is	Download	Weights	Compute
`f16`	Full quality (f16) — Transformer in f16 storage with f32 compute — the largest download, and the reference the other quants here are judged against.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	—
`f16+dit8`	Half size (int8 transformer) — Transformer stored as int8 and computed in f32: roughly half its f16 download, with the execution path left unchanged.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	—
`f16+dit8-a8`	Half size, int8 linear — The int8 transformer with per-token int8 activations in its linear layers — faster on GPUs with dp4a, same download.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8`
`f16+dit8-a8-attn8`	Half size, int8 linear and attention — Adds int8 activations inside attention on top of the int8 linear path; same weights, one more integer stage per step.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8`
`f16+dit8-a8-attn8-s16` (default)	Balanced (int8) — The int8 linear and attention path with attention scores held in f16 — the fastest of the int8 seats here, at f16-level image quality.	3.24 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `i8` / `vae_decoder` = `f16`	`linearCompute` = `a8` / `attentionCompute` = `a8` / `attentionScoreStorage` = `f16`
`f16-c16`	Full quality, f16 compute — f16 storage computed in f16 throughout. Needs the shader-f16 GPU feature, and trades numerical headroom for speed.	5.06 GiB (1.16 GiB shared)	`text_encoder` = `f16` / `text_conditioner` = `f16` / `transformer` = `f16` / `vae_decoder` = `f16`	`linearCompute` = `f16` / `attentionCompute` = `f16` / requires `shaderF16`

Defaults

Any knob not passed to generate() is filled in from the manifest's defaults.

steps: 30
guidanceScale: 4
resolution: 1024 × 1024
negativePrompt: low quality, worst quality, blurry, bad anatomy, jpeg artifacts

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for hdae/karume-anima

Base model

nvidia/Cosmos-Predict2-2B-Text2Image

Finetuned

circlestone-labs/Anima

Finetuned

circlestone-labs/Anima-Base-v1.0-Diffusers

Quantized

(3)

this model

Anima — Karume

What is this

Models and their origins

anima-turbo-v1.1 — Anima Turbo v1.1

anima-v1.0 — Anima Base v1.0

anima-aesthetic-v1.1 — Anima Aesthetic v1.1

anima-turbo-v1.0 — Anima Turbo v1.0

anima-aesthetic-v1.0 — Anima Aesthetic v1.0

License

Models

Usage

Model: anima-turbo-v1.1

Quants

Defaults

Model: anima-v1.0

Quants

Defaults

Model: anima-aesthetic-v1.1

Quants

Defaults

Model: anima-turbo-v1.0

Quants

Defaults

Model: anima-aesthetic-v1.0

Quants

Defaults

Model tree for hdae/karume-anima

`anima-turbo-v1.1` — Anima Turbo v1.1

`anima-v1.0` — Anima Base v1.0

`anima-aesthetic-v1.1` — Anima Aesthetic v1.1

`anima-turbo-v1.0` — Anima Turbo v1.0

`anima-aesthetic-v1.0` — Anima Aesthetic v1.0