GitHub·· 6 天前AI 评分22
Nexus-LLM：终端里的开源 LLM 框架，支持 39+ 模型训练、对话与部署
brutal-45/Nexus-LLM
AI 导读
Nexus-LLM 是一个开源终端 LLM 框架，自带 FastAPI 后端、无需云 API，覆盖模型加载、推理、训练、微调、RAG、智能体与部署全流程，支持 GPT-2、LLaMA、Phi、Qwen 等 39+ 模型。它提供 25+ 斜杠命令、LoRA/QLoRA 微调、FAISS 向量检索、REST 与 WebSocket 服务，并支持 4-bit/8-bit 量化与 Docker 部署。
正文
A Terminal-Based LLM Framework — Train, Chat, Serve, and Deploy Local Language Models
Nexus-LLM is a production-grade, open-source terminal-based LLM framework that works like Claude AI in your terminal — with its own backend, no cloud APIs required. It unifies the entire LLM lifecycle: from model loading and inference to training, fine-tuning, RAG, agents, and deployment. Built with modularity at its core, Nexus-LLM supports 39+ models across all major families.
🌟 Features
Core Engine
- Terminal Chat Interface — Claude AI-like experience in your terminal with streaming, markdown rendering, and syntax highlighting
- Own Backend — No cloud APIs. Runs locally with FastAPI server (REST + WebSocket)
- 39+ Supported Models — GPT-2, GPT-Neo, LLaMA, Phi, Pythia, BLOOM, OPT, Qwen, TinyLlama, CodeGen, and more
- Streaming Inference — Real-time token streaming with multiple sampling strategies
- 25+ Slash Commands —
/help,/model,/switch,/config,/save,/load,/history,/search, and more - Multiple Sampling — Greedy, top-k, top-p (nucleus), typical, eta, epsilon, beam search
Training & Fine-Tuning
- LoRA / QLoRA Fine-Tuning — Parameter-efficient fine-tuning with configurable rank, alpha, and target modules
- Full Training Loop — Gradient accumulation, mixed precision (fp16/bf16), gradient clipping
- 6 LR Schedulers — Linear, cosine, cosine with restarts, polynomial, constant, inverse sqrt
- Curriculum Learning — Progressive difficulty scheduling for improved convergence
- Distributed Training — DDP and FSDP support for multi-GPU training
- Model Export — Export to HuggingFace, ONNX, GGML, safetensors, TorchScript formats
RAG & Knowledge
- Built-in RAG Pipeline — End-to-end retrieve-augment-generate with query expansion
- FAISS Vector Store — Cosine similarity and L2 distance search
- Hybrid Retrieval — BM25 + dense retrieval with reciprocal rank fusion
- Multiple Chunking Strategies — Fixed-size, sentence, paragraph, and semantic chunking
- Embedding Models — SentenceTransformers and HuggingFace embeddings
- Result Reranking — Cross-encoder, MMR diversity, and multi-signal reranking
Agents & Tools
- Agent Framework — Think/Act/Observe loop with 5 agent types (chat, tool, code, research, base)
- Built-in Tools — Calculator, web search, file manager, code runner, weather, note taker, system monitor
- Task Planner — Decompose tasks, create execution plans, manage subtasks
- Agent Memory — Short-term (LRU + TTL), long-term (persistent), and episodic memory
API & Serving
- REST API — Full-featured FastAPI server with
/v1/generate,/v1/chat,/v1/models,/v1/health - WebSocket API — Low-latency bidirectional streaming for interactive applications
- Authentication — API key and token-based auth with role-based permissions
- Rate Limiting — Token bucket and sliding window rate limiting
- Model Serving — Dedicated model server with load balancing and batch processing
- Batched Inference — Efficient batch processing for high-throughput workloads
Safety & Content
- Content Filtering — Keyword, regex, and category-based filtering with PII redaction
- Content Moderation — Severity-based action policies with audit logging
- Toxicity Detection — 9 toxicity categories with weighted compound scoring
- Safety Guardrails — Input/output validation, topic restrictions, prompt injection detection
- Configurable Policies — Default, strict, and permissive policy presets
Infrastructure
- Quantization — 4-bit/8-bit quantization with bitsandbytes, GPTQ, AWQ, and FP8 support
- KV Cache — PagedAttention-style cache management with eviction policies
- CPU Offloading — Layer-wise CPU and disk offloading with prefetching
- Model Adapters — LoRA adapter loading, switching, merging, and multi-adapter serving
- Plugin System — Extensible architecture with lifecycle hooks, 7 built-in plugins
- Docker Support — Multi-stage Docker build with optional CUDA support
- Internationalization — English, Chinese, Japanese, and Spanish localizations
- Monitoring — Terminal dashboard, alerts, resource tracking, and metrics
- Experiment Tracking — MLflow-style experiment management with hyperparameter search and ablation studies
🚀 Quick Start
Installation
# Clone the repository git clone https://github.com/brutal-45/Nexus-LLM.git cd Nexus-LLM # Create virtual environment python -m venv venv source venv/bin/activate # Linux/Mac # venv\Scripts\activate # Windows # Install dependencies pip install -r requirements.txt # Or use the install script chmod +x scripts/install.sh ./scripts/install.sh
Chat (Terminal)
# Start interactive chat with default model (GPT-2 Medium) python main.py chat # Use a specific model python main.py chat --model gpt2-large # List available models python main.py models # Download a model python main.py download --model gpt2-medium
Using the CLI
# After pip install -e . nexus-llm # Start interactive chat nexus-llm chat # Same as above nexus-llm serve # Start API server nexus-llm models # List available models nexus-llm info # Show system info nexus-llm train -m gpt2-medium -d data/train.jsonl
Server
# Start API server python main.py serve --host 0.0.0.0 --port 8000 # Or use the runner script python run_server.py --model gpt2-medium --port 8000 # Docker docker-compose up
Training
# LoRA fine-tuning python main.py train \ --model gpt2-medium \ --data data/training_data.jsonl \ --method lora \ --lora-rank 16 \ --lora-alpha 32 \ --epochs 3 \ --learning-rate 2e-4 # Full training python run_train.py \ --model gpt2-medium \ --data data/training_data.jsonl \ --epochs 5 \ --batch-size 4
API Usage
import requests # Chat completion response = requests.post("http://localhost:8000/v1/chat", json={ "messages": [ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "What is the capital of France?"} ], "temperature": 0.7, "max_tokens": 512 }) print(response.json())
🤖 Supported Models (39+)
| Category | Models | Sizes |
|---|---|---|
| GPT-2 | gpt2, gpt2-medium, gpt2-large, gpt2-xl | 124M - 1.5B |
| GPT-Neo | gpt-neo-125M, gpt-neo-1.3B, gpt-neo-2.7B, gpt-j-6B | 125M - 6B |
| LLaMA | Llama-2-7b, Llama-2-13b | 7B - 13B |
| Phi | phi-1, phi-1.5, phi-2 | 1.3B - 2.7B |
| Pythia | pythia-70m, pythia-160m, pythia-410m, pythia-1b | 70M - 1B |
| BLOOM | bloom-560m, bloom-1b1, bloom-1b7, bloom-3b | 560M - 3B |
| OPT | opt-125m, opt-350m, opt-1.3b, opt-2.7b | 125M - 2.7B |
| Chat | DialoGPT-medium, DialoGPT-large, Qwen1.5-0.5B, Qwen1.5-1.8B | 355M - 1.8B |
| Tiny | TinyLlama-1.1B-Chat-v1.0, llama-2-7b-finetuned | 1.1B - 7B |
| Code | tiny_starcoder_py, codegen-350M-mono, codegen-2B-mono | 164M - 2B |
📁 Project Structure
Nexus-LLM/
├── main.py # CLI entry point (Click)
├── setup.py # Package setup
├── pyproject.toml # Modern Python config
├── requirements.txt # Core dependencies
├── requirements-dev.txt # Dev dependencies
├── requirements-gpu.txt # GPU dependencies
├── Dockerfile # Multi-stage Docker build
├── docker-compose.yml # Docker Compose config
├── Makefile # Build automation
├── download_model.py # Model downloader (39+ models)
├── run_server.py # Server runner
├── run_chat.py # Chat runner
├── run_train.py # Training runner
├── run_eval.py # Evaluation runner
├── run_benchmark.py # Benchmark runner
├── README.md # This file
├── CHANGELOG.md # Version history
├── CONTRIBUTING.md # Contributing guide
├── LICENSE # MIT License
│
├── nexus_llm/ # Main package
│ ├── __init__.py # Package init with public API
│ ├── __main__.py # python -m nexus_llm entry
│ ├── __version__.py # Version info
│ ├── app.py # Main Application class
│ ├── cli.py # Full Click CLI
│ ├── constants.py # All constants
│ ├── exceptions.py # Custom exceptions
│ ├── types.py # Type definitions
│ ├── enums.py # Enumerations
│ ├── registry.py # Component registry
│ ├── events.py # Event system
│ ├── plugins.py # Plugin interface
│ ├── signals.py # Signal handling
│ ├── context.py # Application context
│ ├── state.py # State management
│ ├── config_loader.py # Configuration loading
│ │
│ ├── backend/ # Backend engine (22 files)
│ │ ├── model_manager.py # Model loading & device management
│ │ ├── inference.py # Generation engine
│ │ ├── tokenizer_utils.py # Tokenization utilities
│ │ ├── server.py # FastAPI REST + WebSocket server
│ │ ├── cache.py # KV cache management
│ │ ├── scheduler.py # Request scheduling & batching
│ │ ├── quantization.py # 4-bit/8-bit quantization
│ │ ├── pipeline.py # Pipeline management
│ │ ├── generation.py # Generation config presets
│ │ ├── sampling.py # 9 sampling strategies
│ │ ├── stopping.py # Stopping criteria
│ │ ├── logits_process.py # 14 logits processors
│ │ ├── beam_search.py # Beam search implementation
│ │ ├── streamer.py # Output streaming
│ │ ├── adapter.py # LoRA adapter management
│ │ ├── loader.py # Model loading (HF, safetensors)
│ │ ├── offload.py # CPU/disk offloading
│ │ ├── memory.py # Memory management
│ │ ├── benchmark.py # Backend benchmarking
│ │ ├── health.py # Health checks
│ │ └── metrics.py # Prometheus-style metrics
│ │
│ ├── terminal/ # Terminal UI (21 files)
│ │ ├── chat.py # Interactive chat loop
│ │ ├── formatter.py # Rich output formatter
│ │ ├── commands.py # 25+ slash commands
│ │ ├── history.py # Session persistence
│ │ ├── themes.py # Terminal themes
│ │ ├── prompts.py # Input prompts
│ │ ├── renderer.py # Text renderer
│ │ ├── syntax.py # Syntax highlighting
│ │ ├── markdown_ext.py # Markdown extensions
│ │ ├── widgets.py # UI widgets
│ │ ├── layout.py # Layout manager
│ │ ├── progress.py # Progress indicators
│ │ ├── spinner.py # Loading spinners
│ │ ├── table.py # Table display
│ │ ├── panel.py # Panel display
│ │ ├── status.py # Status bar
│ │ ├── autocomplete.py # Tab completion
│ │ ├── keybinds.py # Key bindings
│ │ ├── multiline.py # Multi-line input
│ │ └── ansi.py # ANSI utilities
│ │
│ ├── training/ # Training pipeline (17 files)
│ │ ├── trainer.py # Training loop
│ │ ├── dataset.py # Dataset handling
│ │ ├── fine_tune.py # LoRA/PEFT fine-tuning
│ │ ├── collator.py # Data collation
│ │ ├── scheduler.py # LR schedulers
│ │ ├── callbacks.py # Training callbacks
│ │ ├── metrics.py # Training metrics
│ │ ├── checkpoint.py # Checkpointing
│ │ ├── distributed.py # Distributed training
│ │ ├── augmentation.py # Data augmentation
│ │ ├── preprocessing.py # Data preprocessing
│ │ ├── evaluation.py # Training evaluation
│ │ ├── optimizer.py # Optimizer configs
│ │ ├── loss.py # Loss functions
│ │ ├── curriculum.py # Curriculum learning
│ │ └── export.py # Model export
│ │
│ ├── config/ # Configuration (10 files)
│ │ ├── settings.py # Settings loader
│ │ ├── model_config.yaml # Model configuration
│ │ ├── training_config.yaml # Training configuration
│ │ ├── server_config.yaml # Server configuration
│ │ ├── ui_config.yaml # UI configuration
│ │ ├── defaults.py # Default values
│ │ ├── validators.py # Config validation
│ │ ├── schema.py # Config schemas
│ │ └── profiles.py # Config profiles
│ │
│ ├── utils/ # Utilities (18 files)
│ │ ├── logger.py # Logging system
│ │ ├── helpers.py # MODEL_CATALOG (39 models)
│ │ ├── io.py # I/O utilities
│ │ ├── system.py # System information
│ │ ├── network.py # Network utilities
│ │ ├── crypto.py # Hash & verification
│ │ ├── text.py # Text processing
│ │ ├── validation.py # Input validation
│ │ ├── decorators.py # Python decorators
│ │ ├── retry.py # Retry logic
│ │ ├── threading.py # Threading utilities
│ │ ├── process.py # Process management
│ │ ├── timer.py # Timing utilities
│ │ ├── hash.py # Hashing utilities
│ │ ├── format.py # Format conversion
│ │ ├── sanitize.py # Input sanitization
│ │ └── profiler.py # Profiling utilities
│ │
│ ├── models/ # Model management (14 files)
│ │ ├── registry.py # Model registry
│ │ ├── catalog.py # 39+ model catalog
│ │ ├── base.py # BaseModel abstract class
│ │ ├── causal_lm.py # CausalLM implementation
│ │ ├── seq2seq.py # Seq2Seq implementation
│ │ ├── chat_model.py # Chat model wrapper
│ │ ├── code_model.py # Code model wrapper
│ │ ├── adapter_model.py # Adapter model (PEFT)
│ │ ├── download.py # Model downloader
│ │ ├── convert.py # Model conversion
│ │ ├── verify.py # Model verification
│ │ ├── benchmark.py # Model benchmarking
│ │ └── quantize.py # Model quantization
│ │
│ ├── api/ # API layer (10 files)
│ │ ├── routes.py # REST API routes
│ │ ├── websocket.py # WebSocket handlers
│ │ ├── middleware.py # API middleware
│ │ ├── auth.py # Authentication
│ │ ├── rate_limit.py # Rate limiting
│ │ ├── schemas.py # Pydantic schemas
│ │ ├── errors.py # Error handling
│ │ ├── cors.py # CORS configuration
│ │ └── docs.py # API documentation
│ │
│ ├── safety/ # Safety module (6 files)
│ │ ├── content_filter.py # Content filtering
│ │ ├── moderation.py # Content moderation
│ │ ├── toxicity.py # Toxicity detection
│ │ ├── guardrails.py # Safety guardrails
│ │ └── policies.py # Safety policies
│ │
│ ├── rag/ # RAG pipeline (8 files)
│ │ ├── retriever.py # Document retrieval
│ │ ├── indexer.py # Document indexing
│ │ ├── embeddings.py # Embedding models
│ │ ├── vector_store.py # FAISS vector store
│ │ ├── chunker.py # Text chunking
│ │ ├── reranker.py # Result reranking
│ │ └── pipeline.py # RAG pipeline
│ │
│ ├── agents/ # Agent framework (10 files)
│ │ ├── base.py # Base agent
│ │ ├── chat_agent.py # Chat agent
│ │ ├── tool_agent.py # Tool-using agent
│ │ ├── code_agent.py # Code agent
│ │ ├── research_agent.py # Research agent
│ │ ├── tools.py # Built-in tools
│ │ ├── planner.py # Task planner
│ │ ├── memory.py # Agent memory
│ │ └── executor.py # Action executor
│ │
│ ├── plugins/ # Plugin system (15 files)
│ │ ├── manager.py # Plugin manager
│ │ ├── loader.py # Plugin loader
│ │ ├── hook.py # Hook system
│ │ ├── builtin/ # Built-in plugins
│ │ │ ├── weather.py # Weather plugin
│ │ │ ├── calculator.py # Calculator plugin
│ │ │ ├── web_search.py # Web search plugin
│ │ │ ├── file_manager.py # File manager plugin
│ │ │ ├── code_runner.py # Code runner plugin
│ │ │ ├── note_taker.py # Note taker plugin
│ │ │ └── system_monitor.py # System monitor plugin
│ │ └── examples/ # Example plugins
│ │ ├── custom_greet.py # Custom greeting
│ │ └── echo.py # Echo plugin
│ │
│ ├── evaluation/ # Evaluation (7 files)
│ │ ├── evaluator.py # Main evaluator
│ │ ├── benchmarks.py # Benchmark runner
│ │ ├── metrics.py # BLEU, ROUGE, F1, etc.
│ │ ├── perplexity.py # Perplexity calculator
│ │ ├── generation_eval.py # Generation quality
│ │ └── report.py # Evaluation reports
│ │
│ ├── data/ # Data handling (7 files)
│ │ ├── loader.py # Data loading
│ │ ├── processor.py # Data processing
│ │ ├── splitter.py # Data splitting
│ │ ├── tokenizer_data.py # Tokenizer data
│ │ ├── converter.py # Format conversion
│ │ └── validator.py # Data validation
│ │
│ ├── templates/ # Templates (13 files)
│ │ ├── prompts/ # Prompt templates
│ │ │ ├── chat.yaml # Chat templates
│ │ │ ├── code.yaml # Code templates
│ │ │ ├── summary.yaml # Summary templates
│ │ │ ├── translate.yaml # Translation templates
│ │ │ ├── creative.yaml # Creative templates
│ │ │ ├── analysis.yaml # Analysis templates
│ │ │ └── default.yaml # Default template
│ │ └── system/ # System prompts
│ │ ├── default.txt # Default system prompt
│ │ ├── coding.txt # Coding assistant
│ │ ├── creative.txt # Creative writing
│ │ ├── assistant.txt # General assistant
│ │ └── researcher.txt # Research assistant
│ │
│ ├── storage/ # Storage (5 files)
│ │ ├── database.py # SQLite database
│ │ ├── conversation_store.py # Conversation persistence
│ │ ├── model_store.py # Model metadata
│ │ └── cache_store.py # Response cache
│ │
│ ├── monitoring/ # Monitoring (5 files)
│ │ ├── dashboard.py # Terminal dashboard
│ │ ├── alerts.py # Alert system
│ │ ├── tracker.py # Resource tracking
│ │ └── reporter.py # Status reporting
│ │
│ ├── multimodal/ # Multimodal (6 files)
│ │ ├── image_processor.py # Image processing
│ │ ├── audio_processor.py # Audio processing
│ │ ├── document_processor.py # Document processing
│ │ ├── vision_model.py # Vision model wrapper
│ │ └── ocr.py # OCR text extraction
│ │
│ ├── serving/ # Model serving (5 files)
│ │ ├── model_server.py # Model server
│ │ ├── load_balancer.py # Load balancing
│ │ ├── queue_manager.py # Request queue
│ │ └── batch_processor.py # Batch processing
│ │
│ ├── experiments/ # Experiments (6 files)
│ │ ├── experiment.py # Experiment tracking
│ │ ├── tracker.py # ML experiment tracker
│ │ ├── comparison.py # Experiment comparison
│ │ ├── hyperparams.py # Hyperparameter search
│ │ └── ablation.py # Ablation studies
│ │
│ ├── migrations/ # Database migrations (5 files)
│ │ ├── migrate.py # Migration runner
│ │ ├── v001_initial.py # Initial schema
│ │ ├── v002_conversations.py # Conversations table
│ │ └── v002_training_jobs.py # Training jobs table
│ │
│ ├── presets/ # Presets (5 files)
│ │ ├── preset_manager.py # Preset manager
│ │ ├── chat_presets.yaml # Chat presets
│ │ ├── training_presets.yaml # Training presets
│ │ └── server_presets.yaml # Server presets
│ │
│ ├── i18n/ # Internationalization (6 files)
│ │ ├── localizer.py # Localization manager
│ │ ├── en.yaml # English
│ │ ├── zh.yaml # Chinese
│ │ ├── ja.yaml # Japanese
│ │ └── es.yaml # Spanish
│ │
│ ├── auth/ # Authentication (4 files)
│ │ ├── manager.py # Auth manager
│ │ ├── tokens.py # Token management
│ │ └── permissions.py # Permission system
│ │
│ ├── cli_ext/ # CLI extensions (4 files)
│ │ ├── completions.py # Shell completions
│ │ ├── formatters.py # Output formatters
│ │ └── validators.py # CLI validators
│ │
│ └── workers/ # Worker processes (5 files)
│ ├── inference_worker.py # Inference worker
│ ├── training_worker.py # Training worker
│ ├── worker_pool.py # Worker pool
│ └── task_queue.py # Task queue
│
├── tests/ # Test suite (160+ files)
│ ├── conftest.py # Shared fixtures
│ ├── test_*.py # Unit & integration tests
│ └── ... # Covering all modules
│
├── scripts/ # Shell scripts (8 files)
│ ├── install.sh # Installation
│ ├── run.sh # Run server
│ ├── train.sh # Training
│ ├── benchmark.sh # Benchmarking
│ ├── setup_env.sh # Environment setup
│ ├── download_models.sh # Model download
│ ├── clean.sh # Cleanup
│ └── update.sh # Update
│
├── data/ # Sample data
│ ├── training_data.jsonl # Training examples
│ ├── eval_data.jsonl # Evaluation examples
│ ├── sample_conversations.json # Sample conversations
│ ├── prompt_templates.json # Prompt templates
│ └── datasets/ # Format examples
│ ├── alpaca_format.jsonl # Alpaca format
│ ├── chat_format.jsonl # Chat format
│ └── instruction_format.jsonl # Instruction format
│
├── docs/ # Documentation (18 files)
│ ├── api/ # API documentation
│ ├── guides/ # User guides
│ ├── tutorials/ # Tutorials
│ └── architecture/ # Architecture docs
│
├── .github/ # GitHub CI/CD
│ ├── workflows/ # GitHub Actions
│ ├── ISSUE_TEMPLATE/ # Issue templates
│ ├── dependabot.yml # Dependabot config
│ └── PULL_REQUEST_TEMPLATE.md # PR template
│
├── .gitignore # Git ignore rules
├── .env # Environment config
├── .env.example # Environment example
├── .pre-commit-config.yaml # Pre-commit hooks
├── .flake8 # Flake8 config
├── .isort.cfg # Isort config
├── .mypy.ini # MyPy config
├── .pylintrc # Pylint config
├── codecov.yml # Codecov config
├── tox.ini # Tox config
├── MANIFEST.in # Package manifest
└── setup.cfg # Setup configuration
⚙️ Configuration
Nexus-LLM can be configured via environment variables, YAML config files, or CLI arguments.
Environment Variables
# Core NEXUS_MODEL=gpt2-medium NEXUS_HOST=0.0.0.0 NEXUS_PORT=8000 NEXUS_DEVICE=auto # auto, cpu, cuda, mps # Generation NEXUS_TEMPERATURE=0.7 NEXUS_TOP_P=0.9 NEXUS_TOP_K=50 NEXUS_MAX_TOKENS=512 # Training NEXUS_LEARNING_RATE=2e-4 NEXUS_BATCH_SIZE=4 NEXUS_EPOCHS=3 # Logging NEXUS_LOG_LEVEL=INFO
Config File (YAML)
# nexus_llm/config/model_config.yaml model: name: gpt2-medium device: auto max_length: 1024 generation: temperature: 0.7 top_p: 0.9 top_k: 50 max_new_tokens: 512 repetition_penalty: 1.1 server: host: 0.0.0.0 port: 8000 workers: 1 training: learning_rate: 2e-4 batch_size: 4 epochs: 3 method: lora lora_rank: 16 lora_alpha: 32
🔌 Plugin Development
Create custom plugins by implementing the PluginInterface:
from nexus_llm.plugins import PluginInterface class MyPlugin(PluginInterface): name = "my_plugin" version = "1.0.0" description = "My custom plugin" def on_load(self): print("Plugin loaded!") def on_unload(self): print("Plugin unloaded!") def on_message(self, message): # Process messages return message
🧪 Testing
# Run all tests pytest tests/ # Run with coverage pytest tests/ --cov=nexus_llm --cov-report=html # Run specific module tests pytest tests/test_inference.py -v # Using Make make test
🐳 Docker
# Build image docker build -t nexus-llm . # Run container docker run -p 8000:8000 nexus-llm # With GPU docker run --gpus all -p 8000:8000 nexus-llm # Using Docker Compose docker-compose up
🤝 Contributing
We welcome contributions! Please see our Contributing Guide for details.
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
📊 Stats
| Metric | Count |
|---|---|
| Source Files | 495+ |
| Python Modules | 20+ |
| Test Files | 160+ |
| Supported Models | 39+ |
| Slash Commands | 25+ |
| Built-in Plugins | 7 |
| Documentation Pages | 18 |
| Shell Scripts | 8 |
📄 License
This project is licensed under the MIT License — see the LICENSE file for details.
Built with ❤️ by the Nexus-LLM Team
来源：GitHub · github.com