Text Generation
Transformers
Safetensors
qwen2
coder
code
agent
conversational
text-generation-inference
Instructions to use AdminReal/NexusCoder with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use AdminReal/NexusCoder with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="AdminReal/NexusCoder") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("AdminReal/NexusCoder") model = AutoModelForCausalLM.from_pretrained("AdminReal/NexusCoder", device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=40) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use AdminReal/NexusCoder with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "AdminReal/NexusCoder" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AdminReal/NexusCoder", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/AdminReal/NexusCoder
- SGLang
How to use AdminReal/NexusCoder with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "AdminReal/NexusCoder" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AdminReal/NexusCoder", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "AdminReal/NexusCoder" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AdminReal/NexusCoder", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use AdminReal/NexusCoder with Docker Model Runner:
docker model run hf.co/AdminReal/NexusCoder
File size: 6,600 Bytes
eca5751 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 | ============================================================================
NEXUS CODER v0.4 — CYBERFORGE EDITION
AI Code & Security Engine — Open Source
============================================================================
Created by: Hieu Louis
GitHub: https://github.com/mhieuhonda/NexusCoder
Year: 2026
License: NAL-1.0 (Attribution Required)
Version: 0.4.0
Python: 3.12.13
┌──────────────────────────────────────────────────────────────────────────┐
│ │
│ NEXUS CODER — SIÊU AI MÃ NGUỒN MỞ CHO CODE & BẢO MẬT │
│ │
│ • 423 tỷ tham số tổng, 39 tỷ tham số kích hoạt mỗi token │
│ • Cửa sổ ngữ cảnh 3 TRIỆU tokens │
│ • 60+ kỹ năng (skills) tích hợp │
│ • 80+ công cụ (tools) tự động đăng ký │
│ • Kiến trúc MoE Transformer thế hệ mới │
│ │
└──────────────────────────────────────────────────────────────────────────┘
TẠI SAO NEXUS CODER KHÁC BIỆT?
==============================
Nexus Coder v0.4 là một kiến trúc AI mã nguồn mở hoàn chỉnh, được Hieu Louis
thiết kế từ con số không. Repository này cung cấp:
✓ Toàn bộ mã nguồn kiến trúc model (Python/PyTorch)
✓ Pipeline thu thập và xử lý dữ liệu code từ hàng nghìn GitHub repos
✓ Framework huấn luyện đa giai đoạn
✓ 60+ skills (code generation, debugging, security audit, ...)
✓ 80+ tools (file ops, exec, web, database, devops, ...)
✓ Tương thích Python 3.12.13 (strict)
TRUNG THỰC VỀ TRẠNG THÁI MODEL
================================
⚠ REPO NÀY KHÔNG CHỨA MODEL ĐÃ ĐƯỢC TRAIN.
Nexus Coder v0.4 phân phối MÃ NGUỒN của kiến trúc, pipeline dữ liệu,
và framework huấn luyện. Người dùng tự huấn luyện mô hình trên dữ
liệu của mình. Mọi thông tin quảng cáo về "performance" hay "benchmark"
chỉ là ước tính lý thuyết dựa trên kích thước kiến trúc — chưa có
model thực tế nào được train và đánh giá chính thức.
Khi bạn thấy ai đó chia sẻ "Nexus Coder đã đạt X điểm benchmark Y", hãy
hỏi xem họ có train model thực tế hay không, và với dữ liệu gì.
TÍNH NĂNG KỸ THUẬT CHÍNH
==========================
• Kiến trúc MoE Transformer với GQA (Grouped Query Attention)
• RoPE + YaRN scaling cho context window cực dài (3M tokens)
• FlashAttention-2 + SDPA + manual fallback
• Sliding Window Attention cho long-context efficiency
• QK-norm (Llama-3 style) cho training stability
• KV cache quantization (int8 / fp8) cho inference memory
• MLP-parallel (gate + up fuses thành 1 matmul)
• Gradient checkpointing cho training VRAM tiết kiệm
• Adaptive Density Routing (top-2 → top-8 experts theo input)
• 48 experts chuyên biệt hóa theo domain code (Python, JS, Rust, ...)
• 8 nguồn dữ liệu: GitHub curated corpus (1000+ repos), HuggingFace,
arXiv, Wikipedia, StackOverflow, The-Stack v2, StarCoder2-data,
Python-Alpaca
• Tích hợp 5 framework tham chiếu: litgpt, LlamaFactory, axolotl,
OpenHands, omp-gym (xem ATTRIBUTIONS.md)
CẤU HÌNH VARIANTS
=================
tiny — 5M params (CPU demo)
small — 125M params (1 GPU)
medium — 1B params (4-8 GPU)
large — 10B params (32+ GPU) — backward-compat với v0.3
xlarge — ~30B params (64+ GPU)
30b — 30B/3B (64-128 GPU, H100 cluster)
70b — ~70B/~12B (research only)
423b — 423B/39B + 3M context (DEFAULT v0.4) — frontier scale
CÀI ĐẶT
========
git clone https://github.com/mhieuhonda/NexusCoder.git
cd NexusCoder
python3.12.13 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
SỬ DỤNG
========
# Xem tóm tắt cấu hình
python -c "from nexus.config import print_config_summary; print_config_summary()"
# Tiny demo
python scripts/train.py --config tiny --steps 100
# Train (cần GPU)
python scripts/train.py --config large --steps 5000 --use-amp
# Chat với Nexus Agent
python scripts/chat.py
GIẤY PHÉP — NAL-1.0 (ATTRIBUTION REQUIRED)
==========================================
Nexus Coder v0.4 được phát hành dưới giấy phép NexusCoder Attribution
License v1.0 (NAL-1.0). Bạn được phép:
✓ Sử dụng cho bất kỳ mục đích nào (commercial hoặc non-commercial)
✓ Sửa đổi, phân phối, sublicense
✓ Train, fine-tune, distill, quantize, ...
✓ Build sản phẩm, dịch vụ, nghiên cứu trên nền Nexus Coder
BẮT BUỘC:
• Phải ghi danh tác giả gốc: "Hieu Louis"
• Phải kèm link: https://github.com/mhieuhonda/NexusCoder
• Trong model cards, README, UI, About pages, API responses,
research citations — bất cứ nơi nào hợp lý và thông dụng.
Không được:
✗ Xóa hoặc làm mờ attribution notices
✗ Cầm quyền tác giả của người khác
✗ Implement technical measures để erase embedded authorship
Xem LICENSE để biết chi tiết đầy đủ.
TÁC GIẢ
========
Hieu Louis — 2026
GitHub: https://github.com/mhieuhonda
Project: https://github.com/mhieuhonda/NexusCoder
License: NAL-1.0 (Attribution Required)
KẾT LUẬN
========
Nexus Coder v0.4 là một dự án mã nguồn mở cá nhân, được xây dựng với
tinh thần minh bạch: đây là CODE của kiến trúc AI, KHÔNG PHẢI model đã
train. Bạn có thể dùng nó để tự train model của mình, học hỏi kiến trúc,
hoặc build sản phẩm riêng. Hãy ghi danh Hieu Louis khi làm vậy.
Cảm ơn bạn đã quan tâm đến Nexus Coder.
— Hieu Louis, 2026
|