Magnus — SYNAXIM .symb Format (INT4)

Powered by SYNAXIM — Symbiotic Native Axiom Inference Machine
Framework-Free LLM Inference | Attention ≡ Memory | O(1) State

---

⚡ v0.1.x is text-only. This release supports text generation, chat, and streaming. Multimodal support (vision, audio) is planned for v0.2.x.

About Magnus

Magnus is a custom fine-tuned Mistral-7B-Instruct-v0.3 with axim-alignment — stripped to text-only for maximum inference speed through SYNAXIM's O(1) Symbiotic State Engine.

• No KV-Cache — O(1) fixed-size memory matrix
• No PyTorch required — pure NumPy core (+ Numba CPU / Triton GPU acceleration)
• Proprietary .symb format — INT4 bitpacked, 3.8× compression
• Text-only — optimized for text generation, chat, and streaming

Quick Start

pip install git+https://github.com/GRRN-MAKER/SYNAXIM.git
huggingface-cli download GRRNMAKE/Magnus-SYMB --local-dir ./magnus-symb

from grrn_inference import GRRNModel

model = GRRNModel.from_pretrained("./magnus-symb", device="cuda")  # or "cpu"

result = model.generate("The meaning of life is", max_tokens=50, temperature=0.7)
print(result.text)

Model Details

Property	Value
Name	Magnus
Modality	Text-only (v0.1.x)
Base Model	mistralai/Mistral-7B-Instruct-v0.3
Parameters	~7.24B
Hidden Size	4096
Layers	32
Attention	32Q / 8KV (GQA 4:1)
Vocabulary	32,768
RoPE θ	1,000,000
Format	.symb INT4 (group_size=128)
Size	~3.85 GB
Adapter	32-layer Symbiotic Gate (knowledge distillation)

Device Selection

model = GRRNModel.from_pretrained("./magnus-symb", device="cpu")       # Auto (Numba if available)
model = GRRNModel.from_pretrained("./magnus-symb", device="cuda")      # Triton GPU kernels
model = GRRNModel.from_pretrained("./magnus-symb", device="cpu-numpy") # Pure NumPy

Backend	Speed	Hardware
cuda	🔥 Fastest	NVIDIA RTX / H100 / A100 (Triton fused kernels)
cpu	⚡ Fast	Any CPU (Numba LLVM → AVX-512/NEON SIMD)
cpu-numpy	Baseline	Any CPU (NumPy only)

Roadmap

Feature	Status
Text generation / chat / streaming	✅ Done
.symb INT4 format	✅ Done
CPU engine (NumPy + Numba)	✅ Done
GPU engine (Triton fused kernels)	✅ Done
Symbiotic Gate adapter training	✅ Done
Magnus as default model	✅ Done
OpenAI-compatible API server	✅ Done
Vision / image support	🔜 Planned (v0.2.x)
Audio support	🔜 Planned (v0.2.x)
Multi-GPU tensor parallelism	🔜 Planned
Rust CPU backend	🔜 Planned

Links

• Engine: github.com/GRRN-MAKER/SYNAXIM
• Model (original): GRRNMAKE/Magnus
• Model (.symb): GRRNMAKE/Magnus-SYMB
• Author: GRRNMAKER

---

SYNAXIM — Because inference should be a machine, not a framework.
Built by GRRNMAKER