Update README to match Qwen2.5-Coder based Nexus Coder

README.md

@@ -1,134 +1,78 @@
+---
+license: apache-2.0
+base_model: Qwen/Qwen2.5-Coder-1.5B-Instruct
+tags:
+  - qwen2
+  - coder
+  - code
+  - agent
+  - transformers
+  - text-generation
+pipeline_tag: text-generation
+library_name: transformers
+---
+
 <div align="center">
 
 # 🧠 Nexus Coder
 
-### AI Code & Security Engine — CyberForge Edition
-
-**An open architecture for next‑generation code generation and security analysis**
-
-[![Python](https://img.shields.io/badge/Python-3.12.13-blue.svg)](https://www.python.org/)
-[![PyTorch](https://img.shields.io/badge/PyTorch-2.0+-ee4c2c.svg)](https://pytorch.org/)
-[![License: NAL-1.0](https://img.shields.io/badge/License-NAL--1.0-orange.svg)](LICENSE)
-[![Status: In Development](https://img.shields.io/badge/Status-In%20Development-yellow.svg)]()
-[![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)]()
-[![GitHub stars](https://img.shields.io/github/stars/mhieuhonda/NexusCoder?style=social)](https://github.com/mhieuhonda/NexusCoder)
-[![GitHub forks](https://img.shields.io/github/forks/mhieuhonda/NexusCoder?style=social)](https://github.com/mhieuhonda/NexusCoder)
-[![GitHub last commit](https://img.shields.io/github/last-commit/mhieuhonda/NexusCoder)](https://github.com/mhieuhonda/NexusCoder)
+### AI Code & Security Engine — Qwen2.5-Coder based
 
-**Created by [Hieu Louis](https://github.com/mhieuhonda)** · 2026
+**A deployable coding model: Qwen2.5-Coder-1.5B weights + Nexus Coder agent engine (skills & tools)**
 
 </div>
 
 ## 📖 Introduction
 
-**Nexus Coder** is an open‑source AI architecture, designed from the ground up by **Hieu Louis**, focused on two core capabilities:
-
-- **High‑quality code generation** powered by a large‑scale Mixture‑of‑Experts (MoE) Transformer.
-- **Deep security analysis** for source code and systems.
-
-The project is under **active development**. This repository provides:
+**Nexus Coder** is a working AI model combining:
 
-- The complete **model architecture source code** (Python/PyTorch).
-- A **data collection and processing pipeline** for code from multiple sources.
-- A **multi‑stage training framework** designed to scale.
-- **60+ skills** and **80+ tools** with automatic registration.
-- Configurations ranging from `tiny` (5M) to `423b` (423B parameters).
+- **Model weights:** [Qwen2.5-Coder-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-Coder-1.5B-Instruct) (Apache 2.0) — a strong 1.5B coding model.
+- **Agent engine:** the Nexus Coder framework (`nexus/`) — 60+ skills and 80+ tools with automatic registration, agent planner/router/memory/safety, data pipeline, and training utilities.
 
-> **Important:** The model is **not pretrained** yet. We distribute only the architecture source and training pipeline. Users need to train their own models on their own data, in compliance with the NAL‑1.0 license.
+## 🚀 Quick Usage
 
-## 📊 Key Technical Specifications
+The model is a standard `Qwen2ForCausalLM`. Load it with `transformers`:
 
-| Item | Value |
-|------|-------|
-| Total parameters | ~423B |
-| Active parameters per token | ~39B |
-| Context window | 3,000,000 tokens (3M) |
-| Architecture | MoE Transformer (GQA + RoPE/YaRN + RMSNorm + SwiGLU + FlashAttention‑2 + Sliding Window + QK‑norm + KV cache quantization + MLP‑parallel + Gradient checkpointing) |
-| Skills | 60+ (code, devops, ML, data, security, cloud, system, blockchain, language) |
-| Tools | 80+ (file, exec, web, code analysis, database, devops, crypto, math, network) |
-| Data sources | 8+ (GitHub curated corpus, HuggingFace, arXiv, Wikipedia, StackOverflow, The‑Stack v2, StarCoder2‑data, Python‑Alpaca) |
-| Python version | 3.12.13 (strict) |
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-## 🚀 Quick Install
+model_id = "AdminReal/NexusCoder"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id)
 
-```bash
-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
-# or: pip install -e ".[all]"
+messages = [{"role": "system", "content": "You are Nexus Coder, a helpful coding assistant."},
+            {"role": "user", "content": "Write a Python function to compute fibonacci."}]
+text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+inputs = tokenizer([text], return_tensors="pt")
+out = model.generate(**inputs, max_new_tokens=512)
+print(tokenizer.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
 ```
 
-💻 Usage
+CLI chat demo (from source):
 
 ```bash
-# Print configuration summary
-python -c "from nexus.config import print_config_summary; print_config_summary()"
-
-# Tiny demo (CPU)
-python scripts/train.py --config tiny --steps 100
-
-# Train larger configurations (requires GPU)
-python scripts/train.py --config large --steps 5000 --use-amp
-python scripts/train.py --config 423b --steps 50000 --use-amp --deepspeed
+python scripts/chat.py --model AdminReal/NexusCoder
 ```
 
-📁 Project Structure
-
-```
-NexusCoder/
-├── nexus/                 # Main package
-│   ├── model/             # MoE Transformer (attention, MoE, layers, ...)
-│   ├── tokenizer/
-│   ├── training/          # Trainer + Dataset
-│   ├── inference/
-│   ├── agent/             # Planner, Router, Memory, Safety
-│   ├── skills/            # 60+ skills (auto‑discovery)
-│   ├── tools/             # 80+ tools (auto‑discovery)
-│   ├── data/              # Collectors + Processors
-│   ├── optim/             # Quantize, LoRA, Distill, Prune
-│   ├── safety/            # Filters, Guardrails
-│   ├── eval/              # Benchmarks, Metrics
-│   ├── integrations/      # litgpt, LlamaFactory, axolotl, OpenHands, omp‑gym
-│   └── utils/
-├── configs/               # YAML configs (tiny → 423B)
-├── scripts/              # CLI scripts
-├── docs/                 # ARCHITECTURE, TRAINING, SKILLS, TOOLS, DATA
-├── tests/
-├── ATTRIBUTIONS.md
-├── CHANGELOG.md
-├── LICENSE                # NAL‑1.0 (Attribution Required)
-├── requirements.txt
-├── pyproject.toml
-├── setup.py
-└── README.md
-```
+## 📁 Repository Contents
 
-⚖️ License
+| Part | What it is | License |
+|------|-----------|---------|
+| `model.safetensors`, `config.json`, `tokenizer.*` | Qwen2.5-Coder-1.5B-Instruct weights | Apache 2.0 |
+| `nexus/` | Agent engine source (skills, tools, agent, data, optim) | NAL-1.0 |
+| `configs/` | Experimental architecture designs (`tiny` → `423b`) — **not** the hosted model | NAL-1.0 |
+| `docs/`, `scripts/`, `tests/` | Documentation, CLI scripts, tests | NAL-1.0 |
 
-Released under the NexusCoder Attribution License v1.0 (NAL‑1.0).
+> The `configs/nexus_coder_*.yaml` files describe a from-scratch MoE research architecture and are **independent** from the Qwen2-based weights hosted in this repo.
 
-· You may use, modify, distribute, and train models for any purpose.
-· Attribution is required to the original author: Hieu Louis (github.com/mhieuhonda).
-· No warranty. See LICENSE for details.
+## ⚖️ Licenses
 
-👤 Author
+- **Model weights:** Apache 2.0 (from Qwen/Qwen2.5-Coder-1.5B-Instruct). See `LICENSE`.
+- **Source code (nexus engine):** NexusCoder Attribution License v1.0 (NAL-1.0). Attribution required to Hieu Louis (github.com/mhieuhonda).
 
-<div align="center">
+## 👤 Author
 
 Hieu Louis · 2026
 
-· GitHub: @mhieuhonda
-· Project: NexusCoder
-· License: NAL‑1.0 (Attribution Required)
-
-</div>
-
-<div align="center">
-
-Nexus Coder — CyberForge Edition
-
-Made by Hieu Louis · 2026
-
-</div>
+- GitHub: @mhieuhonda
+- Project: NexusCoder