Package Exports

@c0mpute/worker
@c0mpute/worker/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@c0mpute/worker) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

c0mpute-worker

Native CLI worker for the c0mpute.ai distributed inference network. Runs LLM inference using node-llama-cpp with full GPU acceleration (CUDA, Metal, Vulkan) and connects to the orchestrator via Socket.io.

Quick Start

npx c0mpute-worker --token <your-token>

What It Does

Detects your GPU (CUDA, Metal, Vulkan, or CPU fallback)
Downloads the optimal GGUF model for your hardware (~8GB)
Runs a speed benchmark
Connects to the c0mpute.ai orchestrator
Accepts and processes inference jobs, streaming tokens back in real time

Options

--token <token>   Authentication token from c0mpute.ai (required)
--url <url>       Orchestrator URL (default: https://c0mpute.ai)
--model <path>    Path to a custom GGUF model file
--benchmark       Run benchmark only, then exit
--version         Show version
--help            Show help

Requirements

Node.js 18+
10GB+ disk space for model download
GPU with 10GB+ VRAM recommended (NVIDIA, Apple Silicon, or Vulkan-compatible)
CPU-only mode available but slower

Default Model

Qwen2.5-14B-Instruct (Q4_K_M quantization) from bartowski/Qwen2.5-14B-Instruct-GGUF.

Models are stored in ~/.c0mpute/models/.

License

MIT