darksidewalker/MiniMaxH3

Official repository for the GRAVITON weight-space fusion architecture and merged MiniMax H3 audio-video DiT checkpoints.

• GRAVITON Architecture, Mathematical Specification & Engine Author: Darksidewalker
• Official Reference Implementation: darksidewalker/dasiwa-quant-station (core/model_merge_engine.py, commit 69412c0+)
• Engine & Workbench License: GPL-3.0-or-later
• Specification & Whitepaper License: CC-BY-SA-4.0 (Attribution + ShareAlike required)

---

Evolution of MiniMax H3 Merge Architectures

This repository encompasses three generations of model-level H3 fusion recipes implemented in DaSiWa Quant Station. While earlier approaches targeted specific conditioning unifications (fl2va + ref2va), GRAVITON—designed and engineered by Darksidewalker—provides a mathematically rigorous, calibration-free spectral and polar geodesic framework for arbitrary DiT pairings.

Recipe	Engine ID	Primary Use Case	Core Mechanism	Key Limitation Solved by Next Generation
1. Hybrid H3	h3_hybrid	Grafting ref2va conditioning onto fl2va	Hard block/row splicing of AdaLN weights & biases (default blocks 30–49) + optional final_layer.audio_out swap.	Zero interpolation; hard boundaries between block regimes; requires unquantized matching layouts.
2. Delta-Fused H3	h3_delta	Single checkpoint serving both keyframe & reference paths	Linear delta injection (W = fl2va + s · Δ) with optional fixed-rank (N) randomized SVD truncation on 2-D trunk matrices.	Fixed rank ignores aspect ratios; linear interpolation on heterogeneous finetunes/turbos causes variance collapse.
3. GRAVITON	h3_graviton	Merging any two H3 models (Base, Turbo, Finetunes, W6A8)	Full-spectrum RMT filtering (Gavish-Donoho) + pre-geodesic TIES/EMR sign mediation + row-wise polar SLERP & RegMean norm re-anchoring.	Current architecture — prevents variance drift, spectral tail noise, post-SLERP trajectory warping, and AdaLN black-frame collapse.

---

1. GRAVITON (h3_graviton)

Geodesic Rank-Adaptive Variance-Isolated Tensor Orthogonal Normalization
Designed and authored by Darksidewalker.

Authorship & Architectural Lineage

While originally motivated by an unimplemented, four-step community recipe description (OliviaRossi/Minimax-h3_Singularity-x-10Eros-Max_PRISM, which had no published code or formal paper), GRAVITON is an original mathematical formulation and systems implementation by Darksidewalker (core/model_merge_engine.py, commit 69412c0+) that corrects fundamental geometric flaws in prior descriptions and introduces:

1. Pre-Geodesic Sign Mediation Ordering: Moves TIES/EMR sign-conflict resolution before spherical interpolation so post-rotation sign flips never knock rows off the geodesic hypersphere.
2. Exact Full-Spectrum RMT & Gavish-Donoho Auto-Rank: Replaces arbitrary fixed-rank cutoffs with aspect-ratio-adaptive optimal hard thresholding via exact Gram-matrix eigendecomposition.
3. Analytical Marchenko-Pastur Quantization Floor (tau_quant) & Comfy-Kitchen W6A8 Bridge: Derives exact uniform step-noise variance from QuantStation W6A8 scale companions and blocks quantization grid staircases from entering the structural subspace ($L$) or diffuse residual ($R$).
4. AdaLN Anti-Alignment Protection & Structural Preservation: Prevents the black-video variance collapse caused by per-row SLERP on anti-aligned AdaLN modulation weights or mismatched community coordinate tables.
5. Decoupled Float Output: Emits clean floating-point artifacts (BF16/F16/F32) rather than risking inline quantization packer corruption.

Why GRAVITON?

Merging step-distilled accelerators (8-step Turbo/DMD/LCM), diverse community finetunes, or group-quantized checkpoints into base DiT models triggers four primary failure modes:

1. Spectral Tail Contamination: Step distillation concentrates structural trajectory changes into top singular vectors while leaving discretization artifacts in the spectral tail. Fixed-rank truncation (e.g., rank-64) under-fits wide layers and leaks noise from narrow projections.
2. Post-Geodesic Trajectory Warping: Applying element-wise sign mediation after spherical interpolation flips components off the hypersphere and destroys the interpolated angle.
3. Timestep Conditioning Instability: Scaling deltas by α > 1.0 or applying per-row SLERP inside adaptive layer norm (adaln) weights causes severe variance drift across denoising steps, resulting in saturated or black-video outputs.
4. Quantized Manifold Mismatch: Group-quantized checkpoints carry Hadamard-rotated (ConvRot), bit-packed coordinates whose discretization noise inflates the delta spectrum.

Mathematical Pipeline

Given base weights $A \in \mathbb{R}^{m \times n}$ and target weights $B \in \mathbb{R}^{m \times n}$ with merge strength $t \in (0, 1]$ (extrapolation $t > 1$ is strictly refused):

Stage 0 — Comfy-Kitchen W6A8 Ingestion & Dequantization Bridge (Opt-In)

Fail-closed by default (accepting only F16/BF16/F32). When allow_quantized_inputs is enabled, the engine recognizes QuantStation's exact W6A8 header contract (I8 packed .weight, F8E4M3 .weight_s_rel, F32 .weight_s_channel, and per-layer .comfy_quant JSON) and dequantizes via comfy_kitchen's AsymW4A8Int8Layout.dequantize—cleanly inverting the ConvRot block-Hadamard rotation ($Q = Q^T = H_b / \sqrt{b}$). From the scale tensors $S = s_{\text{channel}} \cdot s_{\text{rel}}$, the analytical per-element uniform step noise variance is recorded: $${\sigma }_q^2=\frac{\text{mean}(S^2)}{12}\backslash sigma\_q^2\; =\; \backslash frac\{\backslash text\{mean\}(S^2)\}\{12\}$$

Stage 1 — Full-Spectrum RMT Split with Analytical Quantization Floor

For SVD-eligible 2-D trunk matrices, the weight delta $\Delta = B - A$ is decomposed via the exact spectrum of the smaller Gram matrix ($G = \Delta^T \Delta = V \Lambda V^T$, $\sigma_i = \sqrt{\max(\lambda_i, 0)}$): $$\omega (\beta )=0.56{\beta }^3-0.95{\beta }^2+1.82\beta +1.43,\beta =\frac{\min (m,n)}{\max (m,n)}\backslash omega(\backslash beta)\; =\; 0.56\backslash beta^3\; -\; 0.95\backslash beta^2\; +\; 1.82\backslash beta\; +\; 1.43,\; \backslash quad\; \backslash beta\; =\; \backslash frac\{\backslash min(m,n)\}\{\backslash max(m,n)\}$$ $${\tau }_{\text{GD}}=\omega (\beta )\cdot {\sigma }_{\text{med}}\text{(Gavish-Donoho optimal hard threshold)}\backslash tau\_\{\backslash text\{GD\}\}\; =\; \backslash omega(\backslash beta)\; \backslash cdot\; \backslash sigma\_\{\backslash text\{med\}\}\; \backslash quad\; \backslash text\{(Gavish-Donoho\; optimal\; hard\; threshold)\}$$ $${\tau }_{\text{quant}}=(\sqrt{m}+\sqrt{n})\sqrt{{\sigma }_{qA}^2+{\sigma }_{qB}^2}\text{(Marchenko-Pastur bulk edge)}\backslash tau\_\{\backslash text\{quant\}\}\; =\; (\backslash sqrt\{m\}\; +\; \backslash sqrt\{n\})\backslash sqrt\{\backslash sigma\_\{qA\}^2\; +\; \backslash sigma\_\{qB\}^2\}\; \backslash quad\; \backslash text\{(Marchenko-Pastur\; bulk\; edge)\}$$ $$\tau =\max ({\tau }_{\text{GD}},{\tau }_{\text{quant}})\backslash tau\; =\; \backslash max(\backslash tau\_\{\backslash text\{GD\}\},\; \backslash tau\_\{\backslash text\{quant\}\})$$

Singular vectors above $\tau$ form the low-rank structural update $L = (\Delta V_{\text{keep}})V_{\text{keep}}^T$, leaving the diffuse residual $R = \Delta - L$. The denoised target $B_{\text{raw}}$ is reconstructed with an energy-derived halo damping factor: $$B_{\text{raw}}=A+L+{\gamma }_{\text{halo}}R,{\gamma }_{\text{halo}}=1-0.08\cdot \text{capture}(L)\in [0.92,1.0]B\_\{\backslash text\{raw\}\}\; =\; A\; +\; L\; +\; \backslash gamma\_\{\backslash text\{halo\}\}\; R,\; \backslash quad\; \backslash gamma\_\{\backslash text\{halo\}\}\; =\; 1\; -\; 0.08\; \backslash cdot\; \backslash text\{capture\}(L)\; \backslash in\; [0.92,\; 1.0]$$ Whenever dequantization noise is present, $\gamma_{\text{halo}}$ is strictly capped at 0.15 so quantization grid staircases in $R$ never survive into the merge. (For float pairs, fixed-rank $N$ via Halko et al. randomized SVD remains available as an opt-out).

Stage 2 — Pre-Geodesic Sign & Interference Mediation

Executed on $B_{\text{raw}}$ before rotation: salient elements ($\vert{}B_{\text{raw}} - A\vert{} \ge \text{row median}$) whose sign conflicts with $A$ resolve to the magnitude-weighted winner ($t \cdot \vert{}B_{\text{raw}}\vert{}$ vs. $(1-t) \cdot \vert{}A\vert{}$, following TIES/EMR logic) to yield $B'$. Performing mediation prior to SLERP guarantees the spherical trajectory remains unwarped.

Stage 3 — Polar Geodesic Interpolation (Row-Wise)

Directional rotation (short-path SLERP) is decoupled from log-space geometric magnitude interpolation per output row $i$: $${\widehat{A}}_i=\frac{A_i}{\mathrm{\parallel }{A}_i\mathrm{\parallel }},{\widehat{B}}_i^{\mathrm{\prime }}=\frac{B_i^{\mathrm{\prime }}}{\mathrm{\parallel }{B}_i^{\mathrm{\prime }}\mathrm{\parallel }},{\theta }_i=\arccos (\text{clamp}(⟨{\widehat{A}}_i,{\widehat{B}}_i^{\mathrm{\prime }}⟩,-1+ϵ,1-ϵ))\backslash hat\{A\}\_i\; =\; \backslash frac\{A\_i\}\{\backslash Vert\{\}A\_i\backslash Vert\{\}\},\; \backslash quad\; \backslash hat\{B\}\text{'}\_i\; =\; \backslash frac\{B\text{'}\_i\}\{\backslash Vert\{\}B\text{'}\_i\backslash Vert\{\}\},\; \backslash quad\; \backslash theta\_i\; =\; \backslash arccos(\backslash text\{clamp\}(\backslash langle\; \backslash hat\{A\}\_i,\; \backslash hat\{B\}\text{'}\_i\; \backslash rangle,\; -1+\backslash epsilon,\; 1-\backslash epsilon))$$ $$u_i=\frac{\sin ((1-t){\theta }_i){\widehat{A}}_i+\sin (t{\theta }_i){\widehat{B}}_i^{\mathrm{\prime }}}{\sin ({\theta }_i)}u\_i\; =\; \backslash frac\{\backslash sin((1-t)\backslash theta\_i)\backslash hat\{A\}\_i\; +\; \backslash sin(t\backslash theta\_i)\backslash hat\{B\}\text{'}\_i\}\{\backslash sin(\backslash theta\_i)\}$$ $$W_i=\frac{u_i}{\mathrm{\parallel }{u}_i\mathrm{\parallel }}\exp ((1-t)\ln \mathrm{\parallel }{A}_i\mathrm{\parallel }+t\ln \mathrm{\parallel }{B}_i^{\mathrm{\prime }}\mathrm{\parallel })W\_i\; =\; \backslash frac\{u\_i\}\{\backslash Vert\{\}u\_i\backslash Vert\{\}\}\; \backslash exp\backslash big((1-t)\backslash ln\backslash Vert\{\}A\_i\backslash Vert\{\}\; +\; t\backslash ln\backslash Vert\{\}B\text{'}\_i\backslash Vert\{\}\backslash big)$$ (Falls back to linear interpolation when $\theta_i < 10^{-4}$; negates $\hat{A}_i$ for the short path when $\cos\theta_i < 0$).

Stage 4 — Magnitude Re-Anchor & Decoupled Float Export

Each row $W_i$ is rescaled to the geometric norm target of the original, pre-denoised parent pair $\Vert{}A_i\Vert{}^{1-t} \cdot \Vert{}B_i\Vert{}^t$ (RegMean-style norm preservation). Merged weights are always emitted as clean floating-point tensors (BF16/F16/F32) and handed to downstream conversion jobs if packed W6A8 output is desired, eliminating inline packer corruption.

---

Tensor Family & Structural Preservation Policy

GRAVITON routes parameters strictly by architectural role:

• 2-D Trunk Matrices (attn, mlp, out, proj): Full 4-stage GRAVITON pipeline (Spectral Denoise → Sign Mediation → Polar SLERP → Norm Re-Anchor).
• AdaLN Modulation Weights (ada): Spectral Denoise + purely linear delta (W = A + t · Δ_denoised). Because fl2va and ref2va AdaLN rows are strongly anti-aligned, per-row SLERP induces black-video collapse.
• Exact 1-D & Conditioning Families (adaln_t_table / time_embedder, norms, biases, rope.inv_freq): Exact linear interpolation W = A + t(B - A).
• Structural-Layer Preserve (preserve_structural=True by default): Protects conditioning-critical keys (adaln_t_table, time_embedder.*, adaln_proj.linear.*, per-block/qk norms, rope.inv_freq, final_layer.*, token_refiner.*, and patch/conditioning projections) by copying them directly from Base (A) when merging checkpoints with custom coordinate-system folds.

Recommended Operational Regimes (h3_graviton)

Pairing	Anchor (Base A)	Target (Overlay B)	Recommended $t$	Expected Behavior
Base + Turbo (Balanced)	Base bf16	8-step Turbo	0.45 – 0.55	Preserves base generative diversity with faster step convergence.
Turbo + Base (8-Step Lock)	8-step Turbo	Base bf16	0.20 – 0.30	Locks the 8-step distillation schedule while injecting base textures/detail.
Base + Base (Finetunes)	Finetune 1	Finetune 2	0.30 – 0.70	RMT thresholding strips uncorrelated overfitting noise between finetunes.
Turbo + Turbo	Turbo 1	Turbo 2	0.40 – 0.60	Combines distilled aesthetics with zero AdaLN variance drift ($t \le 1.0$).
Float + W6A8	Base bf16	QuantStation w6a8_g32	0.35 – 0.50	MP quantization floor blocks grid artifacts from entering $L$; outputs clean bf16.

---

2. Earlier Approaches (h3_delta & h3_hybrid)

Delta-Fused MiniMax H3 (h3_delta)

Fuses the full ref2va − fl2va weight delta back into an fl2va base so a single checkpoint serves both keyframe (fl2va) and reference (ref2va) conditioning across both pruned (adaln_t_table) and full (time_embedder) variants: $$W=W_{\text{fl2va}}+s\cdot (W_{\text{ref2va}}-W_{\text{fl2va}})W\; =\; W\_\{\backslash text\{fl2va\}\}\; +\; s\; \backslash cdot\; (W\_\{\backslash text\{ref2va\}\}\; -\; W\_\{\backslash text\{fl2va\}\})$$

• Rank 0 (Exact): Applies the full uncompressed delta across all tensors (~38 GB output).
• Rank N (SVD-Capped): Applies randomized SVD capped at rank $N$ (e.g., $N = 1024$) to 2-D trunk matrices while keeping norms, biases, timestep tables, and RoPE frequencies exact.

Hybrid MiniMax H3 (h3_hybrid)

A zero-interpolation block/modality surgery recipe that grafts ref2va AdaLN modulation onto an fl2va base:

• Whole-Block Mode: Copies entire ref2va AdaLN weights and biases into a selected block range (default blocks 30–49 of the 50-block grid).
• Modality-Row Mode (Experimental): Independently splices ref2va video, audio, and text AdaLN rows across user-selected block grids, with optional final_layer.audio_out weight/bias replacement.

---

License, Copyright & Attribution

Copyright © 2026 Darksidewalker. All rights reserved on original methodology, whitepaper, and merge engine architecture.

This repository operates under a split licensing model to protect original open-source mathematical and software contributions while respecting upstream foundation model weights:

1. GRAVITON Engine & DaSiWa Quant Station Implementation (core/model_merge_engine.py):
   Licensed under the GNU General Public License v3.0 or later (GPL-3.0-or-later). Any software derivative, port, or integration of the merge engine must remain open-source under GPL-3.0-or-later and preserve full copyright and NOTICE attribution to Darksidewalker.
2. GRAVITON Whitepaper, Mathematical Formulation & Documentation:
   Licensed under the Creative Commons Attribution-ShareAlike 4.0 International License (CC-BY-SA-4.0). Any adaptations or derivative specifications must credit Darksidewalker and be distributed under the same license.
3. Merged Model Checkpoints (.safetensors):
   Merged weights are derivative artifacts of the MiniMax H3 base models and remain subject to the upstream MiniMax H3 Community License Agreement (minimax-h3-community-license-agreement, as applied to MiniMaxAI/MiniMax-H3) and its acceptable use policies. When redistributing checkpoints merged with GRAVITON, preserve the .safetensors header provenance keys (minimax_h3_graviton=baked and h3_graviton_* metadata).

Citation

If you use the GRAVITON merge architecture, the analytical W6A8 Marchenko-Pastur quantization floor, or checkpoints from this repository in your research, merge toolkits, or derivative models, please cite:

@software{darksidewalker2026graviton,
  author       = {Darksidewalker},
  title        = {{GRAVITON: Geodesic Rank-Adaptive Variance-Isolated Tensor Orthogonal Normalization (DaSiWa Quant Station)}},
  year         = {2026},
  month        = nov,
  publisher    = {GitHub / Hugging Face},
  license      = {GPL-3.0-or-later},
  url          = {https://github.com/darksidewalker/dasiwa-quant-station},
  howpublished = {\url{https://huggingface.co/darksidewalker/MiniMaxH3}},
  note         = {Reference implementation in dasiwa-quant-station (core/model_merge_engine.py, commit 69412c0)}
}

---

References & Grounded Analogs

• Optimal Singular Value Hard Thresholding: Gavish & Donoho, The Optimal Hard Threshold for Singular Values is $4/\sqrt{3}$ (arXiv:1305.5870)
• Randomized Low-Rank Matrix Approximation: Halko, Martinsson, & Tropp (arXiv:0909.4061)
• DARE (Drop And REscale): Yu et al. (arXiv:2311.03099)
• TIES-Merging & EMR-Merging (Sign Conflict Mediation): Yadav et al. (arXiv:2306.01708) / Huang et al. (arXiv:2405.17461)
• Spherical Linear Interpolation (MergeKit): Goddard et al. (arXiv:2403.13257)
• RegMean (Norm Preservation): Jin et al. (arXiv:2212.09849)
• Upstream Community Inspirations: OliviaRossi/Minimax-h3_Singularity-x-10Eros-Max_PRISM (conceptual inspiration) and diffusers-modular/MiniMax-H3-Pruned-Ref-Delta-Fused-r1024