John S

jam_model_memory. An associative memory system that stores and retrieves experiences using the 5W1H framework (Who, What, When, Where, Why, How) and content-addressable memory.

abliterator. Orthogonal Projection Abliteration toolkit featuring Norm-Preservation, Null-Space Constaints, Winsorization, and Adaptive Layer Weighting

latent_control_adapters. Multi-vector latent space steering adapter module for language models

deep_research_protocol. Local LLM integration with web search and persistent memory via llama.cpp and SearXNG.

lm_world_gen. Generate complete fictional world content (regions, factions, characters, dialog trees) using LLMs.

gemma_feature_studio. Gemma-3 SAE Analysis and Steering System with Neuronpedia explanations built with Gemma-Scope-2, SAE-Lens, and React

lora_craft. An open-source web application for fine-tuning large language models using Low-Rank Adaptation (LoRA) and Group Relative Policy Optimization (GRPO). Built to make fine-tuning accessible!

GoCode. GoCode is an autonomous AI development assistant that runs entirely on your machine. It features automatic llama-server management, intelligent project analysis, hybrid retrieval systems, and a full suite of development tools with configurable human-in-the-loop confirmation.

tome_raider. An LLM powered dataset builder for creating high-quality training datasets for GRPO LoRA fine-tuning. Build professional-grade datasets with data collection, generation, quality control, and transformation capabilities.

ai_job_finder. AI-powered job hunting automation - scrape jobs from Indeed and Glassdoor, then match them to your resume using an LLM.