Alex1343543's picture
Upload README.md with huggingface_hub
831022f verified
|
Raw
History Blame Contribute Delete
6.05 kB
metadata
language: en
tags:
  - video-generation
  - minimax-h3
  - ref2va
  - int8
  - convrot
  - turbo
  - h3ddle
  - pulpcut
license: other
license_name: minimax-h3-community-license
license_link: >-
  https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/939557dc319dd91227e30195a763f272ba7f8765/LICENSE
base_model: MiniMaxAI/MiniMax-H3
pretty_name: PulpCut MiniMax H3 Ref2VA Turbo INT8 ConvRot

MiniMax H3 Ref2VA Turbo · pruned INT8 ConvRot (single file)

What this repository is

A single-file MiniMax H3 Ref2VA diffusion transformer — the omni-reference variant, which conditions generation on ordered reference images — with the lightx2v turbo step-distillation merged into the weights, quantized in the same pruned INT8 ConvRot layout as the Comfy-Org release. It is a drop-in replacement for minimax_h3_ref2va_pruned_int8_convrot.safetensors in any runtime that reads the optimized INT8 single-file layout — including H3ddle, the open-source native macOS app it was built for.

This file is not a standalone model. It needs the rest of the optimized package (Qwen3-VL-32B INT8 text encoder, video/audio VAEs, tokenizer) from Comfy-Org/MiniMax-H3, and the FL2VA transformer alongside it if you also want prompt-only and keyframe generation.

Why this merge was made and republished

Every published turbo LoRA for MiniMax H3 targets the FL2VA transformer. Reference generation therefore had no fast path at all: on Apple silicon a short reference clip costs roughly fifty minutes at the base model's twenty denoising passes, with no distilled alternative to fall back on. Since the two transformers are structurally identical — 932 tensors, identical names and shapes — the FL2VA distillation can be merged into the Ref2VA weights, which is what this file is.

Does the distillation survive the transfer?

The adapter was distilled against FL2VA, and reference-identity transfer is exactly the capability step distillation is most likely to damage, so this was measured rather than assumed. Matched control on an M1 Pro — same prompt, same reference image, same seed, 448², adapter as the only variable:

passes wall time identity vs reference (SSIM)
base Ref2VA 20 49.0 min 0.426
this file 8 20.2 min 0.476

2.4x faster with identity transfer intact. The reference subject's markings, build, and setting carry through; detail in fur and whiskers is slightly sharper than the base at its full pass count. The SSIM difference is modest and comes from a single prompt, so treat it as evidence that identity transfer is not degraded rather than proof that it improves.

Recommended settings: 8 denoising passes, euler sampling, all 50 blocks.

Known behavior and limitations

The pruned ComfyUI conversion of the turbo LoRA drops all 51 AdaLN adapter pairs (the source targets AdaLN input dimension 2688, while pruned "compact-curve" models use dimension 8), and its own metadata warns that four-step distillation behaviour may therefore be degraded. We use 8 passes rather than 4 for that reason. The same caveat applies to our FL2VA turbo file; see ModelTC/Minimax-H3-Turbo#7.

How the merge/quantization is done (high level)

For each of the 208 quantized projections, the BF16 pruned Ref2VA base weight is merged with strength × B·A (rank-64, strength 1.0, ema_pruned variant), rotated by the grouped 256-wide Hadamard transform used by the ConvRot runtime kernels, and requantized with symmetric per-row absmax INT8 scales. Token-refiner adapters merge losslessly in BF16. All other tensors are copied byte-identical from the official INT8 file. The pipeline reproduces the official quantizer exactly: run at strength 0 against the Ref2VA base it regenerates the official file with all 3,046,400 scales identical and 1,711 of 19.27 billion int8 values differing (rounding ties).

Source and attribution

Licensing

Derivative of MiniMax H3 weights; the MiniMax H3 Community License Agreement applies. By downloading you agree to its terms.

What this file is used for in H3ddle

H3ddle installs it as a managed model alongside the reference-capable package: the app verifies the SHA-256 below, reuses the shared package files it already has via hardlinks, and defaults the model to 8 passes. Published by PulpCut, whose editor family shares the local-first media generation stack that H3ddle implements in the open.

Safety and intended use

Intended for local, personal video generation. The merge changes speed characteristics, not the base model's content behavior; all usage restrictions of the MiniMax H3 Community License apply unchanged.

File inventory

File Bytes SHA-256
minimax_h3_ref2va_pruned_turbo_int8_convrot.safetensors 20,970,379,854 e64cef63bc2785bcd72e6103c52aa78c6cd2c4f9870a7ce79675083fd65cf2e7

Reproducibility references

The conversion is a single dependency-free Python script, Scripts/convert-turbo-package.py in the H3ddle repository, including the strength-0 self-check used to validate the pipeline against the official file.

Contact

Open an issue in the H3ddle repository.