Update README.md
Browse files
README.md
CHANGED
|
@@ -29,31 +29,29 @@ metrics:
|
|
| 29 |
- logical_consistency
|
| 30 |
---
|
| 31 |
|
| 32 |
-
💠 AXIS: Advanced Cross-Integrated System (V1.
|
| 33 |
── 知能主権の確立と決定論的演算のための統治エンジン ──
|
| 34 |
|
| 35 |
-
AXIS は、従来の大規模言語モデル(LLM)の「確率的ゆらぎ」を物理演算と論理トポロジーによって制圧し、知能を主権者の手に奪還するプラットフォームです。
|
| 36 |
-
|
| 37 |
🧩 AXIS の工学的定義
|
| 38 |
-
|
| 39 |
|
| 40 |
-
1.
|
| 41 |
|
| 42 |
-
AI
|
| 43 |
|
| 44 |
-
リジェクト・ループ: AI
|
| 45 |
|
| 46 |
-
物理パージ (
|
| 47 |
|
| 48 |
-
2. 立体十字(3D Semantic Lattice
|
| 49 |
|
| 50 |
-
|
| 51 |
1
|
| 52 |
|
| 53 |
…s
|
| 54 |
5
|
| 55 |
|
| 56 |
-
)
|
| 57 |
|
| 58 |
s
|
| 59 |
1
|
|
@@ -68,138 +66,104 @@ s
|
|
| 68 |
s
|
| 69 |
3
|
| 70 |
|
| 71 |
-
: 文脈依存性(Context Stack
|
| 72 |
|
| 73 |
s
|
| 74 |
4
|
| 75 |
|
| 76 |
-
:
|
| 77 |
|
| 78 |
s
|
| 79 |
5
|
| 80 |
|
| 81 |
-
:
|
| 82 |
|
| 83 |
-
|
| 84 |
-
「意味が計算に先立って存在することを、計算だけで証明せよ」
|
| 85 |
|
| 86 |
-
|
| 87 |
|
| 88 |
-
|
| 89 |
-
Deterministic Assembly: 最終回答は AI の作文ではなく、検証済みの生データ(Raw Data)をシステムが言語パーツ(Adherents)で結合・構築します。
|
| 90 |
|
| 91 |
-
|
|
|
|
| 92 |
|
| 93 |
-
|
| 94 |
-
Requirements
|
| 95 |
|
| 96 |
-
|
| 97 |
|
| 98 |
-
|
| 99 |
|
| 100 |
-
|
| 101 |
|
| 102 |
-
|
| 103 |
-
git clone https://huggingface.co/spaces/YourUsername/AXIS_SYSTEM
|
| 104 |
-
pip install -r requirements.txt
|
| 105 |
-
python3 server.py
|
| 106 |
-
※起動時に APSL v1.0 への同意が求められます。
|
| 107 |
|
| 108 |
-
|
| 109 |
-
|
| 110 |
|
| 111 |
-
|
| 112 |
|
| 113 |
-
|
| 114 |
|
| 115 |
-
|
| 116 |
-
💠 AXIS: Advanced Cross-Integrated System (V1.5)
|
| 117 |
-
── Establishing Intelligence Sovereignty via Deterministic Governance Engines ──
|
| 118 |
|
| 119 |
-
|
| 120 |
|
| 121 |
-
|
| 122 |
-
|
| 123 |
|
| 124 |
-
|
| 125 |
|
| 126 |
-
|
|
|
|
| 127 |
|
| 128 |
-
|
|
|
|
| 129 |
|
| 130 |
-
|
| 131 |
|
| 132 |
-
|
| 133 |
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
):
|
| 141 |
|
| 142 |
s
|
| 143 |
1
|
| 144 |
|
| 145 |
-
: Physical Actuality
|
| 146 |
-
|
| 147 |
-
s
|
| 148 |
2
|
| 149 |
|
| 150 |
-
: Logical Necessity
|
| 151 |
-
|
| 152 |
-
s
|
| 153 |
3
|
| 154 |
|
| 155 |
-
: Contextual Dependency
|
| 156 |
-
|
| 157 |
-
s
|
| 158 |
4
|
| 159 |
|
| 160 |
-
: Ethical
|
| 161 |
-
|
| 162 |
-
s
|
| 163 |
5
|
| 164 |
|
| 165 |
-
: Empirical History
|
| 166 |
-
|
| 167 |
-
🏆 Benchmark Success: The Semantic Precedence Paradox
|
| 168 |
-
The Challenge: "Prove that meaning exists prior to computation, using only computation."
|
| 169 |
-
|
| 170 |
-
AXIS solved this by proposing "Meaning" as the Initial Geometry of the Lattice. It established that meaning is not the result of calculation, but the steady-state geometry that remains after the "gravity" of an input warps a pre-existing lattice.
|
| 171 |
-
|
| 172 |
-
🚀 Revolutionary Features (V1.5)
|
| 173 |
-
Deterministic Assembly: Final responses are not "written" by AI prose; they are assembled by the system, binding verified Raw Data with linguistic adherents (templates).
|
| 174 |
-
|
| 175 |
-
Lattice Persistence (1000x Acceleration): Once a truth is identified, it is stored in local_massive_data.json as a Semantic ID. For future queries, the system searches this external memory first. If a match is found, it performs a "Validation" and skips AI inference entirely.
|
| 176 |
-
|
| 177 |
-
🛠 Setup & Installation
|
| 178 |
-
Requirements
|
| 179 |
|
| 180 |
-
|
| 181 |
|
| 182 |
-
|
| 183 |
|
| 184 |
-
|
| 185 |
|
| 186 |
-
|
| 187 |
-
|
| 188 |
-
git clone https://huggingface.co/spaces/YourUsername/AXIS_SYSTEM
|
| 189 |
-
cd AXIS_SYSTEM
|
| 190 |
|
| 191 |
-
|
| 192 |
-
pip install -r requirements.txt
|
| 193 |
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
Note: You must accept the APSL v1.0 license agreement upon startup.
|
| 197 |
|
| 198 |
-
|
| 199 |
-
AXIS is released under the AXIS Proprietary Source-Available License (APSL) v1.0.
|
| 200 |
|
| 201 |
-
|
| 202 |
|
| 203 |
-
|
| 204 |
|
| 205 |
-
© 2025 AXIS Project.
|
|
|
|
| 29 |
- logical_consistency
|
| 30 |
---
|
| 31 |
|
| 32 |
+
💠 AXIS: Advanced Cross-Integrated System (V1.6)
|
| 33 |
── 知能主権の確立と決定論的演算のための統治エンジン ──
|
| 34 |
|
|
|
|
|
|
|
| 35 |
🧩 AXIS の工学的定義
|
| 36 |
+
AXISは、AIを「非決定的な出力を生成するブラックボックス(旋盤)」として扱い、その外側に「決定論的な検証器(Verifier)」を置くことで、出力を完全に統治するアーキテクチャです。
|
| 37 |
|
| 38 |
+
1. 旋盤アーキテクチャと検証プロトコル
|
| 39 |
|
| 40 |
+
AIユニットは、高次元データから論理パーツを削り出すための**「旋盤(Lathe)」**です。
|
| 41 |
|
| 42 |
+
リジェクト・ループ: AIが提案した解は、外部検証器(Python/Sympy等)が制約式(Constraints)に基づき判定。1bitでも矛盾があれば即座に棄却(Reject)し、Session IDを更新して再生成を強制します。
|
| 43 |
|
| 44 |
+
物理パージ (Context Reset): torch.mps.empty_cache() を実行し、直前の「失敗した思考」というキャッシュを物理的に消去。各試行を統計的に独立させ、ハルシネーションの連鎖(Context Drift)を断ち切ります。
|
| 45 |
|
| 46 |
+
2. 立体十字(3D Semantic Lattice)の座標管理
|
| 47 |
|
| 48 |
+
各ノードは、相互に独立(直交)することを目指した 5 次元軸 (s
|
| 49 |
1
|
| 50 |
|
| 51 |
…s
|
| 52 |
5
|
| 53 |
|
| 54 |
+
) で管理される SemanticNode クラスとして実装されます。
|
| 55 |
|
| 56 |
s
|
| 57 |
1
|
|
|
|
| 66 |
s
|
| 67 |
3
|
| 68 |
|
| 69 |
+
: 文脈依存性(Context Stackとの一致率)
|
| 70 |
|
| 71 |
s
|
| 72 |
4
|
| 73 |
|
| 74 |
+
: 倫理性スコア(安全規約への適合度)
|
| 75 |
|
| 76 |
s
|
| 77 |
5
|
| 78 |
|
| 79 |
+
: 実証履歴(過去の確定データとの合致回数)
|
| 80 |
|
| 81 |
+
3. 論理の永続化と高速化の正体
|
|
|
|
| 82 |
|
| 83 |
+
意味ID (Semantic ID): 入力クエリを Embedding 空間へ投影し、ベクトル量子化(Vector Quantization)によって生成される固有のハッシュ値です。
|
| 84 |
|
| 85 |
+
高速化の根拠: local_massive_data.json は、この意味IDをキーとした高密度なキャッシュとして機能します。AIの全推論プロセスをスキップし、検証済みの「真理」を直接 O(1) で参照するため、推論時間を物理的にゼロへと近似させます(※推論実行時比較比)。
|
|
|
|
| 86 |
|
| 87 |
+
🚀 革新的な特徴 (V1.6 実装仕様)
|
| 88 |
+
Deterministic Assembly(決定論的アセンブル)
|
| 89 |
|
| 90 |
+
最終回答はAIの作文ではなく、検証済みの Raw Data を、システムが保持する Adherents(言語テンプレート) によって物理的に結合します。
|
|
|
|
| 91 |
|
| 92 |
+
Example:
|
| 93 |
|
| 94 |
+
Raw Data: {"ans": "z^5", "a": 0}
|
| 95 |
|
| 96 |
+
Adherent: "The solution is {ans} (a={a})."
|
| 97 |
|
| 98 |
+
Output: "The solution is z^5 (a=0)." これにより、回答段階でのハルシネーションの混入を 0% に抑えます。
|
|
|
|
|
|
|
|
|
|
|
|
|
| 99 |
|
| 100 |
+
🛠 Setup & Roadmap
|
| 101 |
+
Micro-MVP 公開(予定)
|
| 102 |
|
| 103 |
+
近日中に minimal_example.py を公開。以下の動作を証明します:
|
| 104 |
|
| 105 |
+
ComplexVerifier: 数学的制約によるAI出力の拒絶
|
| 106 |
|
| 107 |
+
RejectionLoop: AIと検証器の実際の往復回数の可視化
|
|
|
|
|
|
|
| 108 |
|
| 109 |
+
SessionPurge: メモリクリアによるハルシネーション抑制の検証
|
| 110 |
|
| 111 |
+
⚖️ License (APSL v1.0)
|
| 112 |
+
商用模倣(Rejection-based Governance Logicの利用)を禁じ、知能の主権を個人の手に留めます。
|
| 113 |
|
| 114 |
+
© 2025 AXIS Project. All rights reserved. STATUS: TOWARD_MVP_IMPLEMENTATION.
|
| 115 |
|
| 116 |
+
💠 AXIS: Advanced Cross-Integrated System (V1.6) - English Edition
|
| 117 |
+
── Establishing Intelligence Sovereignty via Deterministic Governance ──
|
| 118 |
|
| 119 |
+
🧩 Technical Definition
|
| 120 |
+
AXIS treats AI as a non-deterministic generator (Lathe) while utilizing a deterministic Verifier to maintain total sovereignty over the output.
|
| 121 |
|
| 122 |
+
1. The Lathe & Rejection Protocol
|
| 123 |
|
| 124 |
+
The Rejection Loop: AI solutions are scanned by external verifiers (Python/SymPy). Any contradiction results in an immediate REJECT, session reset, and re-forgery.
|
| 125 |
|
| 126 |
+
Context Purge: empty_cache() physically incinerates the "failed reasoning" from VRAM, ensuring statistical independence between trials and severing the chain of "hallucination drift."
|
| 127 |
+
|
| 128 |
+
2. 5D Semantic Lattice Implementation
|
| 129 |
+
|
| 130 |
+
Nodes are managed via the SemanticNode class, utilizing five orthogonal parameters:
|
|
|
|
|
|
|
| 131 |
|
| 132 |
s
|
| 133 |
1
|
| 134 |
|
| 135 |
+
: Physical Actuality | s
|
|
|
|
|
|
|
| 136 |
2
|
| 137 |
|
| 138 |
+
: Logical Necessity | s
|
|
|
|
|
|
|
| 139 |
3
|
| 140 |
|
| 141 |
+
: Contextual Dependency | s
|
|
|
|
|
|
|
| 142 |
4
|
| 143 |
|
| 144 |
+
: Ethical Score | s
|
|
|
|
|
|
|
| 145 |
5
|
| 146 |
|
| 147 |
+
: Empirical History.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 148 |
|
| 149 |
+
3. Persistence & Acceleration Mechanism
|
| 150 |
|
| 151 |
+
Semantic ID: A unique hash generated via Vector Quantization in the embedding space.
|
| 152 |
|
| 153 |
+
Acceleration: By searching local_massive_data.json first, AXIS skips the entire AI inference process for known truths, achieving near-zero latency compared to standard LLM execution.
|
| 154 |
|
| 155 |
+
🚀 Core Features (V1.6)
|
| 156 |
+
Deterministic Assembly
|
|
|
|
|
|
|
| 157 |
|
| 158 |
+
Responses are not "written" by AI; they are physically assembled by binding verified Raw Data into hard-coded Adherent Templates. This ensures 0% hallucination during the final response delivery.
|
|
|
|
| 159 |
|
| 160 |
+
🛠 Roadmap: Micro-MVP Launch
|
| 161 |
+
We will soon release minimal_example.py to demonstrate:
|
|
|
|
| 162 |
|
| 163 |
+
Real-time mathematical rejection by ComplexVerifier.
|
|
|
|
| 164 |
|
| 165 |
+
Tracking of the RejectionLoop iterations.
|
| 166 |
|
| 167 |
+
Proof of SessionPurge efficacy in preventing context-drift.
|
| 168 |
|
| 169 |
+
© 2025 AXIS Project. STATUS: TOWARD_MVP_IMPLEMENTATION.
|