OliviaRossi/Minimax-H3-Hybrid-DMAD-PDMD
text-to-video model by OliviaRossi
From the publisher
Model card & documentation
Source previewRead the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.
[](https://huggingface.co/OliviaRossi/Minimax-H3-Hybrid-DMAD-PDMD) -orange?style=for-the-badge) [](https://huggingface.co/pdmd2026/pdmd4NFElora) [](https://github.com/comfyanonymous/ComfyUI) 4–8 Step Accelerated Inference • Smooth S-Curve Hybrid Backbone • Native INT8 & W6A8 Quantization Minimax-H3-Hybrid-DMAD-PDMD is an accelerated, multimodal reference-capable 33B video-generation foundation model engineered natively for ComfyUI. It resolves the trade-off in the open-weights ecosystem between MiniMax-H3's FL2VA (high dynamic range, natural motion, but missing multi-reference conditioning) and Ref2VA (reference following, but prone to crushed contrast and dark shadow banding). By executing a Smooth S-Curve Hybrid AdaLN Merge, permanently fusing two distillation LoRAs (0.85× DMAD + 0.333× PDMD), and applying low-rank SVD pruning, this model runs generation workflows in 4 to 8 steps on consumer GPUs. Available in two precision tiers tailored for consumer hardware: $S(t) = t^2(3 - 2t), \quad t \in [0, 1]$ For each transformer block $i \in \{0, \dots, 49\}$ and its AdaLN modulation projection $W^{(i)}$: $ \alpha(i) = \begin{cases} 0.0 & \text{if } i < 20 \\ t^2(3 -…
Read the full model card ↗ · Preview checked 2026-11-12T07:37:21.822Z
Inside the original model card — Document outline
- 🎞️ Minimax-H3-Hybrid-DMAD-PDMD
- The High-Speed, Low-VRAM 33B Foundation Model for MiniMax-H3 Video Generation
- ?? Model Releases & Formats
- ??? Pipeline Architecture
- ?? Technical Highlights
- 1. Dual-Distillation Superposition (4–8 Steps)
- 2. Smooth S-Curve Hybrid AdaLN Injection
- 3. Native Quantization Options
- ?? Mathematical Formulations
- 1. Adaptive S-Curve Block Merging
- 2. Dual-Distillation Weight Superposition
- 3. Orthogonal Hadamard ConvRot Rotation
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Documentation belongs to its respective authors. Reported project/model license: other. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
Model overview
OliviaRossi/Minimax-H3-Hybrid-DMAD-PDMD is a text-to-video model repository published by OliviaRossi on Hugging Face. The source reports the minimax-h3 library.
This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.
Model facts
- Task
- text-to-video
- Library
- minimax-h3
- Recent downloads (30 days)
- 15,848
- Cumulative likes
- 18
- Hugging Face trending score
- 4
- Reported safetensors parameters
- Not reported
- Architecture
- Not reported
- License
- other
- Access
- Not gated by Hugging Face
- Created
- 2026-11-03T14:25:32.000Z
- Last modified
- 2026-11-04T08:24:26.000Z
Compatibility and lineage
Reported languages: en
Reported base models: Comfy-Org/MiniMax-H3, pdmd2026/pdmd_4NFE_lora, ZhengmingYu/DMAD
Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.
Use and evaluate
Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.
No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.
Source and freshness
Source: Hugging Face Hub. Metadata observed 2026-11-14T06:59:49.277Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.
Related models
QuantStack/Wan2.2-T2V-A14B-GGUF
QuantStack/Wan2.2-TI2V-5B-GGUF