Skip to content
Images

MOSS-Image-Base DMD2

8 шагов вместо 50

1 min readupdated

LoRA-адаптер для MOSS-Image-Base от data-archetype, дистиллированный по distribution matching. Заменяет 50 guided steps базовой модели на 8 форвардов при CFG 1 без negative prompt. Меняется только transformer; text encoder Qwen3.5-9B, VAE FLUX.2 small, tokenizer и пайплайн остаются от базы.

Ключевые особенности

  • 8 шагов при CFG 1 — guidance уже дистиллирован, negative prompt не нужен.
  • LoRA rank 128, сила 1; адаптирует только transformer.
  • Совместима с большинством файнтюнов MOSS-Image-Base и другими LoRA — грузить вместе.
  • Около 1 мегапикселя, любое соотношение сторон, которое поддерживает база.
  • Сэмплер lcm: на каждом шаге чистый прогноз и повторный шум; другие сэмплеры дают хуже.
  • Для ComfyUI есть готовый workflow и отдельный LoRA-файл.

Пример

from peft import PeftModel
transformer = PeftModel.from_pretrained(
    transformer, "data-archetype/moss_image_base_dmd2",
    subfolder="transformer"
).get_base_model()
image = pipe(prompt, height=896, width=1152,
             num_inference_steps=8, guidance_scale=1.0).images[0]
Подробнее8 шагов при CFG 1 без negative prompt вместо 50 guided steps базы; LoRA rank 128, сила 1.
Адаптирован только transformer; text encoder Qwen3.5-9B, VAE FLUX.2 small и пайплайн — от MOSS-Image-Base.
В ComfyUI нужен bypass-загрузчик: обычный сливает LoRA в bf16, и большая часть изменений теряется.
Лицензия Apache 2.0, как у базовой модели.
Discuss Telegram 22

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # data-archetype/moss_image_base_dmd2
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download data-archetype/moss_image_base_dmd2 --local-dir ~/models/moss_image_base_dmd2
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to moss_image_base_dmd2
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • SageAttention 3v2.2.0from October 28, 2025

    even faster on RTX 50 / Blackwell

    FP4 attention for Blackwell. Faster than SageAttention 2, but the authors suggest v2 when accuracy matters.

    How to enable: Build from the sageattention3_blackwell folder of the repo; in ComfyUI via KJNodes.

    Earlier generations (2)
    • SageAttention 2++ — Quantised (INT8/FP8) attention with no visible quality loss. The main speed-up for image and video generation.
    • SageAttention 1 — First version: INT8 for Q·K. Still works, but 2++ is faster and more accurate.
  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (3)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
    • xFormers — The older way to speed up attention, works almost everywhere. Sage or Flash are faster where available.
  • cache-dit (DBCache)v1.5.6from November 20, 2026new release

    1.5–3× at nearly the same quality

    Reuses block outputs between diffusion steps. Supports most DiT models.

    How to enable: pip install cache-dit; one line cache_dit.enable_cache(pipe) for diffusers.

    Earlier generations (2)
    • MagCache — Skips steps by magnitude of change. More accurate than TeaCache, no calibration.
    • TeaCache — The first popular diffusion step cache. Works, newer methods are more accurate.
  • Nunchaku (SVDQuant 4-bit)v1.2.1from January 25, 2026

    ≈3× faster and 3.5× less memory

    4-bit versions of popular image models with almost no quality loss (FP4 on RTX 50).

    How to enable: Ready weights for supported models; in ComfyUI use ComfyUI-nunchaku.

  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of moss_image_base_dmd2 (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

data-archetype/moss_image_base_dmd2
Author
data-archetype
Category
Images
Released
Posted

On NeyroLearn

Hugging Face

022

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments