isagellm 0.2.2.7

sageLLM: Modular LLM inference engine for domestic computing power (Huawei Ascend, NVIDIA)

sageLLM

Protocol Compliance (Mandatory)

• MUST follow Protocol v0.1: https://github.com/intellistream/sagellm-docs/blob/main/docs/specs/protocol_v0.1.md
• Any globally shared definitions (fields, error codes, metrics, IDs, schemas) MUST be added to Protocol first.

🚀 Modular LLM Inference Engine for Domestic Computing Power

Ollama-like experience for Chinese hardware ecosystems (Huawei Ascend, NVIDIA)

---

✨ Features

• 🎯 One-Click Install - pip install isagellm gets you started immediately
• 🧠 CPU-First - Default CPU engine, no GPU required
• 🇨🇳 Domestic Hardware - First-class support for Huawei Ascend NPU
• 📊 Observable - Built-in metrics (TTFT, TBT, throughput, KV usage)
• 🧩 Plugin System - Extend with custom backends and engines

📦 Quick Install

# Install sageLLM (CPU-first, no GPU required)
pip install isagellm

# With Control Plane (request routing & scheduling)
pip install 'isagellm[control-plane]'

# With API Gateway (OpenAI-compatible REST API)
pip install 'isagellm[gateway]'

# Full server (Control Plane + Gateway)
pip install 'isagellm[server]'

# With CUDA support
pip install 'isagellm[cuda]'

# All features
pip install 'isagellm[all]'

🚀 Quick Start

CLI (like ollama)

# Show system info
sage-llm info

# Default mode (CPU engine, no GPU required)
sage-llm serve
sage-llm run -p "What is LLM inference?"

# Production mode (requires control-plane)
# pip install 'isagellm[server]'
sage-llm serve --control-plane
sage-llm gateway --port 8080

Python API

import asyncio

from sagellm import BackendConfig, EngineConfig, Request, create_backend, create_engine

# Create CPU backend + engine (no GPU needed)
async def main() -> None:
  backend = create_backend(BackendConfig(kind="cpu", device="cpu"))
  engine = create_engine(
    EngineConfig(kind="cpu", model="sshleifer/tiny-gpt2", model_path="sshleifer/tiny-gpt2"),
    backend,
  )

  await engine.start()
  try:
    request = Request(
      request_id="demo-001",
      prompt="Hello, world!",
      max_tokens=128,
      stream=False,
    )
    response = await engine.execute(request)
    print(response.output_text)
  finally:
    await engine.stop()


asyncio.run(main())

print(f"Response: {response.text}")
print(f"TTFT: {response.metrics.ttft_ms:.2f} ms")
print(f"Throughput: {response.metrics.throughput_tps:.2f} tokens/s")

Configuration

# ~/.sage-llm/config.yaml
backend:
  kind: cpu  # Options: cpu, pytorch-cuda, pytorch-ascend
  device: cpu

engine:
  kind: cpu
  model: sshleifer/tiny-gpt2

control_plane:
  endpoint: "localhost:8080"

📊 Metrics & Validation

sageLLM provides comprehensive performance metrics:

{
  "ttft_ms": 45.2,
  "tbt_ms": 12.5,
  "throughput_tps": 80.0,
  "peak_mem_mb": 24576,
  "kv_used_tokens": 4096,
  "prefix_hit_rate": 0.85
}

Run benchmarks:

sage-llm demo --workload year1 --output metrics.json

🏗️ Architecture

isagellm (umbrella package)
├── isagellm-protocol       # Protocol v0.1 types
│   └── Request, Response, Metrics, Error, StreamEvent
├── isagellm-backend        # Hardware abstraction (L1 - Foundation)
│   └── BackendProvider, CPUBackend, (CUDABackend, AscendBackend)
├── isagellm-comm           # Communication primitives (L2 - Infrastructure)
│   └── Topology, CollectiveOps (all_reduce/gather), P2P (send/recv), Overlap
├── isagellm-kv-cache       # KV cache management (L2 - Optional)
│   └── PrefixCache, MemoryPool, EvictionPolicies, Predictor, KV Transfer
├── isagellm-compression    # Inference acceleration (quantization, sparsity, etc.) (L2 - Optional)
│   └── Quantization, Sparsity, SpeculativeDecoding, Fusion
├── isagellm-core           # Engine core & runtime (L3)
│   └── Config, Engine, Factory, DemoRunner, Adapters (vLLM/LMDeploy)
├── isagellm-control-plane  # Request routing & scheduling (L4 - Optional)
│   └── ControlPlaneManager, Router, Policies, Lifecycle
└── isagellm-gateway        # OpenAI-compatible REST API (L5 - Optional)
    └── FastAPI server, /v1/chat/completions, Session management

🔧 Development

Quick Setup (Development Mode)

# Clone all repositories
./scripts/clone-all-repos.sh

# Install all packages in editable mode
./quickstart.sh

# Open all repos in VS Code Multi-root Workspace
code sagellm.code-workspace

📖 See WORKSPACE_GUIDE.md for Multi-root Workspace usage.

Testing

# Clone and setup
git clone https://github.com/IntelliStream/sagellm.git
cd sagellm
pip install -e ".[dev]"

# Run tests
pytest -v

# Format & lint
ruff format .
ruff check . --fix

# Type check
mypy src/sagellm/

# Verify dependency hierarchy
python scripts/verify_dependencies.py

📖 Development Resources

• DEPLOYMENT_GUIDE.md - 完整部署与配置指南
• TROUBLESHOOTING.md - 故障排查快速参考
• ENVIRONMENT_VARIABLES.md - 环境变量完整参考
• DEVELOPER_GUIDE.md - 开发者指南
• WORKSPACE_GUIDE.md - Multi-root Workspace 使用
• INFERENCE_FLOW.md - 推理流程详解
• PR_CHECKLIST.md - Pull Request 检查清单

---

📚 Documentation Index

用户文档

• 快速开始 - 5 分钟上手
• 部署指南 - 生产环境部署
• 配置参考 - 完整配置选项
• 环境变量 - 环境变量参考
• 故障排查 - 常见问题解决

开发者文档

• 开发指南 - 贡献代码
• 架构设计 - 系统架构
• Workspace 使用 - Multi-root 工作区
• PR 检查清单 - 提交前检查

API 文档

• OpenAI 兼容 API - 参见 sagellm-gateway
• Python API - 参见 API_REFERENCE.md（待补充）

子包文档

• sagellm-protocol - 协议定义

• sagellm-backend - 后端抽象

• sagellm-core - 引擎核心

• sagellm-control-plane - 控制面

• sagellm-gateway - API 网关

• sagellm-benchmark - 基准测试

• DEVELOPER_GUIDE.md - 架构规范与开发指南

• PR_CHECKLIST.md - Pull Request 审查清单

• scripts/verify_dependencies.py - 依赖层次验证

📚 Package Details

Package	PyPI Name	Import Name	Description
sagellm	isagellm	sagellm	Umbrella package (install this)
sagellm-protocol	isagellm-protocol	sagellm_protocol	Protocol v0.1 types
sagellm-core	isagellm-core	sagellm_core	Runtime & config
sagellm-backend	isagellm-backend	sagellm_backend	Hardware abstraction

📄 License

Proprietary - IntelliStream. Internal use only.

---

_{Built with ❤️ by IntelliStream Team for domestic AI infrastructure}