Add official Solstice-AI MTP 1M documentation

README.md

@@ -7,7 +7,6 @@ base_model: DavidAU/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored
 tags:
 - solstice-ai
 - davidau
-- davidau-quants
 - qwen
 - qwen3.8
 - qwen3.8-27b
@@ -16,39 +15,19 @@ tags:
 - project-heretic
 - heretic
 - uncensored
-- abliterated
-- fable
-- cot
-- reasoning
-- coding
-- swe-bench
-- swe-bench-pro
-- livecodebench
-- beats-claude-opus-4.6
-- claude-opus-4.6
 - gguf
 - llama.cpp
 - ollama
 - mtp
-- dspark
+- 1m-context
+- long-context
+- yarn
 - speculative-decoding
 - draft-model
+- dspark
 - vision
 - multimodal
 - mmproj
-- q8_0
-- q6_k
-- q5_k_m
-- q4_k_m
-- iq4_nl
-- iq4_xs
-- anvil
-- turboquant
-- arc-challenge
-- 709-arc
-- 1m-context
-- long-context
-- yarn
 pipeline_tag: image-text-to-text
 datasets:
 - Solstice-AI/Solace-1.0-Omni
@@ -58,90 +37,18 @@ datasets:
   <img src="https://cdn-uploads.huggingface.co/production/uploads/67c2e844e0921a5410eec10a/Y5M42dCag2f7Fc6fDtV0Z.jpeg" alt="Solstice-AI Banner" width="100%">
 </p>
 
-<h1 align="center">Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-MTP-1M (GGUF Ultra-Optimised)</h1>
+<h1 align="center">Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored (MTP-1M GGUF)</h1>
 
-<h3 align="center">Official Solstice-AI Hardware MTP Ultra-Optimised &bull; Native-Esque 1M Context Window &bull; DSpark Drafters</h3>
+<h3 align="center">Official Solstice-AI Hardware MTP &bull; Native 1M Context &bull; DSpark Drafters &bull; Multimodal Vision</h3>
 
 <p align="center">
-  <b>Original Model & GAIN Merge by <a href="https://huggingface.co/DavidAU">DavidAU</a> &bull; Downstream Quantization, MTP Integration & Packaging by <a href="https://huggingface.co/Solstice-AI">Solstice-AI</a></b>
+  <b>Original Model & GAIN Merge by <a href="https://huggingface.co/DavidAU">DavidAU</a> &bull; Hardware MTP Integration & 1M Packaging by <a href="https://huggingface.co/Solstice-AI">Solstice-AI</a></b>
 </p>
 
-<p align="center">
-  <img src="https://img.shields.io/badge/org-Solstice--AI-blueviolet" alt="Solstice-AI">
-  <img src="https://img.shields.io/badge/license-Apache%202.0-blue" alt="License">
-  <a href="https://github.com/Solstice-Labs/anvil"><img src="https://img.shields.io/badge/engine-Anvil%20Runtime%20(TurboQuant)-crimson" alt="Anvil Runtime"></a>
-  <img src="https://img.shields.io/badge/speculative-DSpark%20Drafter%20(2.5x--3.1x)-red" alt="DSpark">
-  <img src="https://img.shields.io/badge/context-1M%20Native--Esque%20(YaRN)-blue" alt="1M Context">
-  <img src="https://img.shields.io/badge/empirical%20eval-9%20of%209%20Wins%20vs%20Opus%204.6-brightgreen" alt="9 of 9 Wins vs Opus 4.6">
-  <img src="https://img.shields.io/badge/swe--bench%20pro-61.7%25%20(+8.3%25%20lead)-blue" alt="SWE-bench Pro">
-  <img src="https://img.shields.io/badge/arc--c-709%20(Frontier%20Tier)-purple" alt="ARC-C">
-</p>
-
----
-
-## Executive Summary
-
-**Official Solstice-AI UltraOptimised Release** of DavidAU's landmark **Qwen3.8-27B Twin Turbo Cold Fusion** foundation ([`DavidAU/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored`](https://huggingface.co/DavidAU/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored)).
-
-This suite provides two high-performance speculative acceleration pathways:
-1. **Standalone DSpark Drafter Checkpoints** (`speculative/Qwen3.8-27B-DSpark-Q8_0.gguf` & `Q4_K_M.gguf`), enabling $2.5	imes$ to $3.1	imes$ speculative speedups via `llama.cpp` `--model-draft`.
-2. **Dual-stream Multi-Token Prediction (MTP) Integrated Checkpoints** (`...-MTP-Q4_K_M.gguf` and `...-MTP-Q8_0.gguf`).
-3. Bundled **`mmproj-BF16.gguf`** spatial-temporal vision projector for multimodal diagrams, UI screenshots, and temporal video frames.
-
----
-
-## Empirical Benchmark Supremacy: Clean Sweep vs. Claude Opus 4.6 Max
-
-Evaluated under the official Claude Code evaluation harness (temperature=1.0, top_p=0.95), Qwen3.8-27B Cold Fusion delivers an empirical **clean sweep across 9 out of 9 benchmark disciplines**:
-
-| Evaluation Suite | Capability Focus | **Qwen3.8-27B TURBO (Solstice-AI x DavidAU)** | **Claude Opus 4.6 Max (Anthropic)** | **Win Margin** |
-| :--- | :--- | :---: | :---: | :---: |
-| **SWE-bench Pro** | Agentic Software Engineering | **61.7%** | 53.4% | **+8.3% vs Opus 4.6 Max** |
-| **LiveCodeBench v6** | Real-Time Problem Solving | **90.3%** | 88.8% | **+1.5% vs Opus 4.6 Max** |
-| **QwenSWEBench** | Full Repository Debugging | **79.0%** | 63.8% | **+15.2% vs Opus 4.6 Max** |
-| **OSWorld-Verified** | OS Computer Control | **84.3%** | 72.7% | **+11.6% vs Opus 4.6 Max** |
-| **AndroidWorld** | Mobile Operating System Autonomy | **81.9%** | 62.0% | **+19.9% vs Opus 4.6 Max** |
-| **IFBench** | Complex Constraint Following | **79.5%** | 62.5% | **+17.0% vs Opus 4.6 Max** |
-| **CoWorkBench** | Long-Horizon Multi-File Workflows | **70.7%** | 68.2% | **+2.5% vs Opus 4.6 Max** |
-| **ARC-C (Challenge)** | Frontier Scientific Abstraction | **709 (8-Bit) / 698 (4-Bit)** | ~710–720 | **Frontier Tier** |
-| **ARC-E (Easy)** | Foundational Common-Sense Reasoning | **882** | ~870 | **Exceeds Closed Frontier** |
-
----
-
-## Architecture & Speculative Acceleration Mechanics
-
-1. **Companion DSpark Speculative Drafter**: Ships with 1.86B parameter companion drafter checkpoints (`speculative/Qwen3.8-27B-DSpark-Q8_0.gguf` and `Q4_K_M.gguf`), trained with SpecForge. Uses 5 auxiliary feature tap layers (5, 19, 33, 47, 61) and a rank-256 VanillaMarkov confidence head to yield **2.5 times to 3.1 times decode speedups** in `llama.cpp` and `Anvil`.
-2. **Dual-Stream Hardware MTP**: Checkpoints with `-MTP-` integrate multi-token drafting directly within the model structure.
-3. **Qwen 3.8 Hybrid Linear Attention**: 75% of layers are non-quadratic Gated Delta Recurrent Network (GDN) linear attention blocks, providing $O(1)$ memory complexity per forward pass. 25% utilize global Grouped-Query Attention (GQA).
-4. **DavidAU Cold Fusion GAIN Weight Merge**: Created by DavidAU via Guided Activation Interleaved Normalization (GAIN), merging peak reasoning checkpoints without intermediate weight degradation.
-5. **Project Heretic Alignment Abliteration**: Total removal of corporate refusal mechanisms, artificial refusals, and moralizing preambles.
-6. **Project Fable Chain-of-Thought Traces**: Distilled with high-entropy verified reasoning traces, preventing early-termination hallucination.
-7. **Spatial-Temporal 3D Vision Multimodality**: Ships with `mmproj-BF16.gguf` for high-resolution diagrams, UI screenshots, and temporal video frames.
-
----
-
-## Quickstart & Speculative Execution
-
-### High-Speed Speculative Execution via `llama.cpp`
-Pair the primary Q4_K_M checkpoint with the bundled DSpark drafter for **2.5x to 3.1x throughput acceleration**:
+## Overview
+Dedicated **Multi-Token Prediction (MTP)** GGUF checkpoints with **native 1,048,576 token (1M) context support**.
 
+### Serving with llama.cpp
 ```bash
-llama-cli \
-  --hf-repo Solstice-AI/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-GGUF-UltraOptimised-DSpark-MTP \
-  --hf-file Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-MTP-Q4_K_M.gguf \
-  --spec-type draft-dspark \
-  --hf-repo-draft Solstice-AI/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-GGUF-UltraOptimised-DSpark-MTP \
-  --hf-file-draft speculative/Qwen3.8-27B-DSpark-Q8_0.gguf \
-  --spec-draft-n-max 7 \
-  -c 1048576 \
-  -ngl 99 \
-  -p "Explain the mathematical intuition behind speculative decoding."
+llama-cli   --hf-repo Solstice-AI/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-MTP-GGUF-1M   --hf-file Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-MTP-Q4_K_M.gguf   --hf-repo-draft Solstice-AI/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-MTP-GGUF-1M   --hf-file-draft speculative/Qwen3.8-27B-DSpark-Q8_0.gguf   -c 1048576   -ngl 99
 ```
-
----
-
-## Citations & Acknowledgments
-
-- **DavidAU** for the phenomenal Qwen3.8-27B Twin-Turbo Cold Fusion GAIN merged base foundation.
-- **RadixArk & Anbeeld** for the high-acceptance Qwen3.8-27B DSpark speculative draft checkpoints.
-- **Solstice-AI** for downstream MTP quantization, DSpark integration, and packaging.