Skip to content
Other

MiniMax-H3-Image-Undistill-Lora

an adapter for fine-tuning MiniMax-H3-Image

1 min readupdated

This model from circlestone-labs turns the distilled MiniMax-H3-Image model into a base-model equivalent, making it a practical starting point for custom training runs. Unlike similar adapters that rely on a discrete schedule, this version was trained with a continuous timestep distribution. Built on 10,000 diverse images captioned by Gemma-4-31b, it works without modifying the underlying architecture. It is a useful tool for anyone doing custom diffusion model training.

Key features

  • Removes distillation to mimic a base model
  • Uses a continuous timestep distribution instead of a discrete schedule
  • Trained on 10,000 diverse captioned images
  • Preserves the original architecture with PDD-head selection

What you can do

  • Use it as a foundation for training new LoRAs and checkpoints
  • Run training workflows with it in diffusion-pipe
Read moreDetails
Training: 15 epochs on 10k images.
Resolution: an equal mix of 512 and 1024 pixels.
Notes: weights are in ComfyUI format; may be incompatible with AI-Toolkit due to its strict reliance on a discrete schedule.
Discuss Telegram 2

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.

    # circlestone-labs/MiniMax-H3-Image-Undistill-Lora
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    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 circlestone-labs/MiniMax-H3-Image-Undistill-Lora --local-dir ~/models/MiniMax-H3-Image-Undistill-Lora
    # The model is in ~/models. How to run it is in the model card (link above).
  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.

  • 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 (2)
    • 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.
  • 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.

MiniMax H3 prompts

All prompts

Ready prompts with the resulting video: copy and try.

Model family

New posts about versions of MiniMax-H3-Image-Undistill-Lora (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

circlestone-labs/MiniMax-H3-Image-Undistill-Lora
Author
circlestone-labs
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

02

Ask AI about this model

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

Comments