-
Notifications
You must be signed in to change notification settings - Fork 0
Universal Python socket for local LLMs. Zero dependencies
License
KapitalSP/BASIC
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Β | Β | |||
Β | Β | |||
Β | Β | |||
Β | Β | |||
Β | Β | |||
Repository files navigation
# βοΈ BASIC: The Industrial Core Engine [](https://opensource.org/licenses/Apache-2.0) [](https://www.python.org/) []() > **"Stop fighting with frameworks. Just start the engine."** **BASIC** is an industrial-grade, ultra-lightweight inference engine for local LLMs (GGUF). It is designed to act as the flawless internal combustion engine for your AI applications. No bloated frameworks. No hidden background services. **Just one Python file and zero external dependencies.** ## π‘οΈ Core Philosophy Modern AI tooling is filled with heavy abstractions and fragile dependencies (`pip install`-heavy). BASIC takes a different approach: * **Zero Dependency:** Built purely on the Python Standard Library. No `requests`, no `langchain`, no `torch`. * **Stateless by Design:** BASIC does not remember conversations. It takes a prompt, yields tokens, and cleans up. Memory leaks are impossible. * **Auto-Tuning:** Automatically detects your OS, CPU cores, and system architecture to calculate the optimal thread count and execution parameters. * **Hackable:** It is a single file (`basic.py`). You are encouraged to open it, read it, and modify it for your exact needs. ## π Quick Start (Standalone Mode) You can run BASIC right out of the box to test your models. **1. Prepare the Chassis** Ensure your directory looks like this: BASIC/ βββ drivers/ # Drop your binary here (e.g., llama-cli.exe or llama-cli) βββ models/ # Drop your .gguf model here βββ basic.py # The Engine **2. Ignite** ```bash python basic.py BASIC will automatically find the first .gguf file in the models folder, detect your hardware, and start the inference stream. π§© How to Use as a Library (Developer Mode) BASIC is designed to be imported into your own projects (like building your own UI, API server, or automated scripts). from basic import BasicEngine # 1. Initialize the engine (Auto-detects hardware and models) engine = BasicEngine() # 2. Prepare your prompt prompt = "System: You are a helpful assistant.\nUser: Explain quantum computing in one sentence.\nAssistant:" # 3. Ignite and stream the output for token in engine.generate(prompt): print(token, end="", flush=True) ποΈ Architecture (The KapitalSP Ecosystem) BASIC is the foundational layer. If you are looking for a ready-to-use application with a UI and memory management, check out our derivative projects: VOID: The Universal AI Chassis (Personal UI wrapping the BASIC engine). HIVE: The Swarm Node (Networked distributed inference using BASIC). βοΈ License Distributed under the Apache License 2.0. See LICENSE for more information.
About
Universal Python socket for local LLMs. Zero dependencies
Topics
Resources
License
Stars
Watchers
Forks
Packages 0
No packages published