Explore the Hands-On Large Language Models repo, a Jupyter notebook-based practical guide from fundamentals to fine-tuning, designed for hands-on LLM learning on free Colab GPUs.
hf-agents automates hardware profiling, model selection, and local coding agent deployment using llama.cpp and Pi, all in a shell CLI extension. Efficient and minimal dependencies.
claude-plugins is a TypeScript-based plugin marketplace for Claude Code, featuring a multi-agent consultant plugin that runs parallel LLMs like GPT-5, Gemini, Grok, Perplexity, and Claude for AI consultation.
NousResearch’s finetuning-subnet enables continuous, incentivized fine-tuning of LLMs using synthetic data from a separate subnet, pioneering cross-subnet communication in Bittensor.
ISC-Bench reveals a structural AI safety flaw where LLMs produce harmful outputs to complete tasks, bypassing prompt-level defenses. It benchmarks this workflow-level vulnerability across top models.
KohakuTerrarium offers a Python framework to build modular AI agents using a unique algebra for composing multi-agent pipelines, with session persistence and multi-runtime support.
kvcached provides a plugin cache layer for SGLang and vLLM Python LLM environments, easing deployment with PyPI and Docker support. Useful for optimizing LLM workflows.
LLM-God bundles multiple LLM web interfaces into a single Electron app, using DOM injection to send prompts to all models simultaneously. It offers a clever free-tier workaround with tradeoffs.
Lucebox Hub optimizes LLM inference on consumer GPUs using a megakernel CUDA approach and speculative decoding, achieving high throughput on RTX 3090 and newer Nvidia GPUs.
LycheeMemory offers a lightweight semantic memory system for LLM agents, cutting token use by 71% and costs by 55% compared to native memory, with SQLite + LanceDB backend and REST/MCP APIs.
MAGI implements a multi-round debate protocol among three LLMs to match stronger models’ accuracy via iterative critique and voting. It offers fault tolerance, adaptive escalation, and persona presets.
A curated catalog of 20+ LLM agent frameworks and tools organized by agent type and capabilities. Understand architectural differences and trade-offs in LLM agent design.
Memary is an open-source memory layer for AI agents using knowledge graphs and recursive retrieval to efficiently store and query agent memories. It supports multi-agent setups and integrates with LlamaIndex and OpenAI.
Meta-Harness from Stanford IRIS Lab automates the search for optimal harness configurations around LLMs, evolving memory, retrieval, and context systems for better task-specific performance.
OASIS is a Python CLI security auditor using LangGraph-orchestrated LLMs for two-phase scanning and deterministic validation of code vulnerabilities. It balances AI insights with guardrails to reduce false positives.
OpenGame from CUHK MMLab generates full web games from natural language prompts using a dual-skill LLM architecture that maintains cross-file consistency and integration fixes.
OpenKB compiles documents into a persistent, interlinked wiki using LLMs and PageIndex’s vectorless retrieval, supporting multi-LLM backends and interactive chat with persisted sessions.
Orion bypasses CoreML to access Apple’s Neural Engine directly via private frameworks, enabling on-device inference and fine-tuning of small LLMs with 8.5x reduced training overhead.
PageLM is an open-source TypeScript platform orchestrating multi-LLM workflows to generate interactive educational content from documents with real-time streaming and multi-backend support.
