sllm

Tiny Model, Big Logic: Diversity-Driven Optimization Elicits Large-Model Reasoning Ability in VibeThinker-1.5B

Free Perplexity-style web search library for local sLLM. Bing + DuckDuckGo → semantic filtering → reranking → LLM-ready prompt with citations.

Advanced local-first RAG system powered by Ollama and LangGraph. Optimized for high-performance sLLM orchestration featuring adaptive intent routing, semantic chunking, intelligent hybrid search (FAISS + BM25), and real-time thought streaming. Includes integrated PDF analysis and secure vector caching.

Train reasoning models on your Mac. GRPO training framework for Apple Silicon with curriculum learning.

Llama-3.2-1B-Instruct model and KorQuAD dataset Finetune project

spacevoice is a system that explains information about a space to help visually impaired people empathize with the space.

A lightweight, <1000-line Python log triage agent for multi-source server logs with optional local SLLM review.

Zero-cloud on-device Korean MCQ generator Android SDK & CoT fine-tuning pipeline based on Google Gemma-2-2B (QuizScore 0.985)

A Nix template for running a Small Large Language Model (sLLM).