How does the Ollama integration work?

It provides native streaming support for Ollama endpoints like generate and chat, using a ParallelDispatcher to handle concurrent requests efficiently through a Python-native interface.

How does the self-improvement loop function?

Py-ACSet logs every interaction to a local DuckDB instance, tracking performance metrics like latency and compression ratios to evolve the system's efficiency over time.

What is an ACSet in the context of Py-ACSet?

An ACSet (Attributed C-Set) is a flexible, categorical data structure that allows you to model complex relationships and attributes, used here to manage and validate LLM request-response patterns.

What is the purpose of GF(3) balancing?

GF(3) logic uses 'trits' (-1, 0, +1) to balance request fanout and coordinate different roles—such as hatchery, coordinator, and generator—within an ergodic system.

Py-ACSet Topological LLM Orchestrator

Name: Py-ACSet Topological LLM Orchestrator
Author: plurigrid

byplurigrid

•

データサイエンスとML

Orchestrates complex local LLM workflows using Python-native Attributed C-Sets and categorical data modeling for self-improving action spaces.

Py-ACSet is a specialized skill that bridges the gap between Category Theory and local AI orchestration. It provides a Python-native implementation of Attributed C-Sets (ACSets) integrated with Ollama, enabling developers to build structured, parallelized LLM action spaces. By utilizing GF(3) load balancing and a self-improvement loop via DuckDB logging, it allows for the creation of 'topological chemputers' where LLM interactions are modeled as categorical morphisms, ensuring high-fidelity state transitions and optimized request fanout performance.

主な機能

01Parallel execution with configurable ThreadPoolExecutor and semaphore-based fanning

02GF(3) triadic load balancing for efficient request distribution and coordination

03Advanced categorical modeling including Galois connections and cobordant phase spaces

048 GitHub stars

05Streaming Ollama integration for high-performance LLM action spaces

06Automated self-improvement loops with compression progress tracking via DuckDB

ユースケース

01Orchestrating complex multi-model LLM workflows with local infrastructure and Ollama

02Implementing categorical database structures for structured data collection and reasoning

03Modeling and verifying high-fidelity state transitions in AI-driven agentic systems

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi py-acset

For use in Claude.ai and ChatGPT

Download Skill